Radar-Chart/topic-modelling-dim-reduction.ipynb

{"nbformat":4,"nbformat_minor":0,"metadata":{"colab":{"name":"topic-modelling-dim-reduction.ipynb","provenance":[{"file_id":"1xhZ9BqjFqsuFboGQGm18mHT_wuy7INpG","timestamp":1614640514817},{"file_id":"1hczBDoe0q4olG8zAMREJLIXosxoCRqTU","timestamp":1614198003093}],"collapsed_sections":[],"toc_visible":true},"kernelspec":{"display_name":"Python 3","name":"python3"},"widgets":{"application/vnd.jupyter.widget-state+json":{"4b0b89039f8840a1a8408945ca005475":{"model_module":"@jupyter-widgets/controls","model_name":"HBoxModel","state":{"_view_name":"HBoxView","_dom_classes":[],"_model_name":"HBoxModel","_view_module":"@jupyter-widgets/controls","_model_module_version":"1.5.0","_view_count":null,"_view_module_version":"1.5.0","box_style":"","layout":"IPY_MODEL_ba6bac176442433a9e40ad20a0a4e081","_model_module":"@jupyter-widgets/controls","children":["IPY_MODEL_ad7a571381454cc6bcf459aff99913c4","IPY_MODEL_07e1a86a3b3c47b4a7991cbe24253c83"]}},"ba6bac176442433a9e40ad20a0a4e081":{"model_module":"@jupyter-widgets/base","model_name":"LayoutModel","state":{"_view_name":"LayoutView","grid_template_rows":null,"right":null,"justify_content":null,"_view_module":"@jupyter-widgets/base","overflow":null,"_model_module_version":"1.2.0","_view_count":null,"flex_flow":null,"width":null,"min_width":null,"border":null,"align_items":null,"bottom":null,"_model_module":"@jupyter-widgets/base","top":null,"grid_column":null,"overflow_y":null,"overflow_x":null,"grid_auto_flow":null,"grid_area":null,"grid_template_columns":null,"flex":null,"_model_name":"LayoutModel","justify_items":null,"grid_row":null,"max_height":null,"align_content":null,"visibility":null,"align_self":null,"height":null,"min_height":null,"padding":null,"grid_auto_rows":null,"grid_gap":null,"max_width":null,"order":null,"_view_module_version":"1.2.0","grid_template_areas":null,"object_position":null,"object_fit":null,"grid_auto_columns":null,"margin":null,"display":null,"left":null}},"ad7a571381454cc6bcf459aff99913c4":{"model_module":"@jupyter-widgets/controls","model_name":"FloatProgressModel","state":{"_view_name":"ProgressView","style":"IPY_MODEL_c8403bd4cc95435dbe6e247461772b38","_dom_classes":[],"description":"100%","_model_name":"FloatProgressModel","bar_style":"success","max":244733649,"_view_module":"@jupyter-widgets/controls","_model_module_version":"1.5.0","value":244733649,"_view_count":null,"_view_module_version":"1.5.0","orientation":"horizontal","min":0,"description_tooltip":null,"_model_module":"@jupyter-widgets/controls","layout":"IPY_MODEL_b870e81c04f242f581fe33ae5708a8b3"}},"07e1a86a3b3c47b4a7991cbe24253c83":{"model_module":"@jupyter-widgets/controls","model_name":"HTMLModel","state":{"_view_name":"HTMLView","style":"IPY_MODEL_6ecd6eacb06e4889bbfb7c947f584816","_dom_classes":[],"description":"","_model_name":"HTMLModel","placeholder":"​","_view_module":"@jupyter-widgets/controls","_model_module_version":"1.5.0","value":" 245M/245M [00:12&lt;00:00, 19.5MB/s]","_view_count":null,"_view_module_version":"1.5.0","description_tooltip":null,"_model_module":"@jupyter-widgets/controls","layout":"IPY_MODEL_8307eaff07bc43c1b1fb033833599da8"}},"c8403bd4cc95435dbe6e247461772b38":{"model_module":"@jupyter-widgets/controls","model_name":"ProgressStyleModel","state":{"_view_name":"StyleView","_model_name":"ProgressStyleModel","description_width":"initial","_view_module":"@jupyter-widgets/base","_model_module_version":"1.5.0","_view_count":null,"_view_module_version":"1.2.0","bar_color":null,"_model_module":"@jupyter-widgets/controls"}},"b870e81c04f242f581fe33ae5708a8b3":{"model_module":"@jupyter-widgets/base","model_name":"LayoutModel","state":{"_view_name":"LayoutView","grid_template_rows":null,"right":null,"justify_content":null,"_view_module":"@jupyter-widgets/base","overflow":null,"_model_module_version":"1.2.0","_view_count":null,"flex_flow":null,"width":null,"min_width":null,"border":null,"align_items":null,"bottom":null,"_model_module":"@jupyter-widgets/base","top":null,"grid_column":null,"overflow_y":null,"overflow_x":null,"grid_auto_flow":null,"grid_area":null,"grid_template_columns":null,"flex":null,"_model_name":"LayoutModel","justify_items":null,"grid_row":null,"max_height":null,"align_content":null,"visibility":null,"align_self":null,"height":null,"min_height":null,"padding":null,"grid_auto_rows":null,"grid_gap":null,"max_width":null,"order":null,"_view_module_version":"1.2.0","grid_template_areas":null,"object_position":null,"object_fit":null,"grid_auto_columns":null,"margin":null,"display":null,"left":null}},"6ecd6eacb06e4889bbfb7c947f584816":{"model_module":"@jupyter-widgets/controls","model_name":"DescriptionStyleModel","state":{"_view_name":"StyleView","_model_name":"DescriptionStyleModel","description_width":"","_view_module":"@jupyter-widgets/base","_model_module_version":"1.5.0","_view_count":null,"_view_module_version":"1.2.0","_model_module":"@jupyter-widgets/controls"}},"8307eaff07bc43c1b1fb033833599da8":{"model_module":"@jupyter-widgets/base","model_name":"LayoutModel","state":{"_view_name":"LayoutView","grid_template_rows":null,"right":null,"justify_content":null,"_view_module":"@jupyter-widgets/base","overflow":null,"_model_module_version":"1.2.0","_view_count":null,"flex_flow":null,"width":null,"min_width":null,"border":null,"align_items":null,"bottom":null,"_model_module":"@jupyter-widgets/base","top":null,"grid_column":null,"overflow_y":null,"overflow_x":null,"grid_auto_flow":null,"grid_area":null,"grid_template_columns":null,"flex":null,"_model_name":"LayoutModel","justify_items":null,"grid_row":null,"max_height":null,"align_content":null,"visibility":null,"align_self":null,"height":null,"min_height":null,"padding":null,"grid_auto_rows":null,"grid_gap":null,"max_width":null,"order":null,"_view_module_version":"1.2.0","grid_template_areas":null,"object_position":null,"object_fit":null,"grid_auto_columns":null,"margin":null,"display":null,"left":null}},"ee2b5c1f3a1e4144827d553bfbecfb50":{"model_module":"@jupyter-widgets/controls","model_name":"HBoxModel","state":{"_view_name":"HBoxView","_dom_classes":[],"_model_name":"HBoxModel","_view_module":"@jupyter-widgets/controls","_model_module_version":"1.5.0","_view_count":null,"_view_module_version":"1.5.0","box_style":"","layout":"IPY_MODEL_2ec2512f308f4c1aae3441191c8b220f","_model_module":"@jupyter-widgets/controls","children":["IPY_MODEL_e5973bf0174c423daf5026098e6393c0","IPY_MODEL_8f670ca1fa264a5884bef98b52ad52b6"]}},"2ec2512f308f4c1aae3441191c8b220f":{"model_module":"@jupyter-widgets/base","model_name":"LayoutModel","state":{"_view_name":"LayoutView","grid_template_rows":null,"right":null,"justify_content":null,"_view_module":"@jupyter-widgets/base","overflow":null,"_model_module_version":"1.2.0","_view_count":null,"flex_flow":null,"width":null,"min_width":null,"border":null,"align_items":null,"bottom":null,"_model_module":"@jupyter-widgets/base","top":null,"grid_column":null,"overflow_y":null,"overflow_x":null,"grid_auto_flow":null,"grid_area":null,"grid_template_columns":null,"flex":null,"_model_name":"LayoutModel","justify_items":null,"grid_row":null,"max_height":null,"align_content":null,"visibility":null,"align_self":null,"height":null,"min_height":null,"padding":null,"grid_auto_rows":null,"grid_gap":null,"max_width":null,"order":null,"_view_module_version":"1.2.0","grid_template_areas":null,"object_position":null,"object_fit":null,"grid_auto_columns":null,"margin":null,"display":null,"left":null}},"e5973bf0174c423daf5026098e6393c0":{"model_module":"@jupyter-widgets/controls","model_name":"FloatProgressModel","state":{"_view_name":"ProgressView","style":"IPY_MODEL_7d889b10ac104d73abbc4349e9d350e3","_dom_classes":[],"description":"Batches: 100%","_model_name":"FloatProgressModel","bar_style":"success","max":3832,"_view_module":"@jupyter-widgets/controls","_model_module_version":"1.5.0","value":3832,"_view_count":null,"_view_module_version":"1.5.0","orientation":"horizontal","min":0,"description_tooltip":null,"_model_module":"@jupyter-widgets/controls","layout":"IPY_MODEL_65f28be5c5ab48c38b00b94726093434"}},"8f670ca1fa264a5884bef98b52ad52b6":{"model_module":"@jupyter-widgets/controls","model_name":"HTMLModel","state":{"_view_name":"HTMLView","style":"IPY_MODEL_8a2efb2cf0314c1e9eb61b70e75fd7e3","_dom_classes":[],"description":"","_model_name":"HTMLModel","placeholder":"​","_view_module":"@jupyter-widgets/controls","_model_module_version":"1.5.0","value":" 3832/3832 [1:07:18&lt;00:00,  1.05s/it]","_view_count":null,"_view_module_version":"1.5.0","description_tooltip":null,"_model_module":"@jupyter-widgets/controls","layout":"IPY_MODEL_efa6b686ff5a4c609612fe0f95241633"}},"7d889b10ac104d73abbc4349e9d350e3":{"model_module":"@jupyter-widgets/controls","model_name":"ProgressStyleModel","state":{"_view_name":"StyleView","_model_name":"ProgressStyleModel","description_width":"initial","_view_module":"@jupyter-widgets/base","_model_module_version":"1.5.0","_view_count":null,"_view_module_version":"1.2.0","bar_color":null,"_model_module":"@jupyter-widgets/controls"}},"65f28be5c5ab48c38b00b94726093434":{"model_module":"@jupyter-widgets/base","model_name":"LayoutModel","state":{"_view_name":"LayoutView","grid_template_rows":null,"right":null,"justify_content":null,"_view_module":"@jupyter-widgets/base","overflow":null,"_model_module_version":"1.2.0","_view_count":null,"flex_flow":null,"width":null,"min_width":null,"border":null,"align_items":null,"bottom":null,"_model_module":"@jupyter-widgets/base","top":null,"grid_column":null,"overflow_y":null,"overflow_x":null,"grid_auto_flow":null,"grid_area":null,"grid_template_columns":null,"flex":null,"_model_name":"LayoutModel","justify_items":null,"grid_row":null,"max_height":null,"align_content":null,"visibility":null,"align_self":null,"height":null,"min_height":null,"padding":null,"grid_auto_rows":null,"grid_gap":null,"max_width":null,"order":null,"_view_module_version":"1.2.0","grid_template_areas":null,"object_position":null,"object_fit":null,"grid_auto_columns":null,"margin":null,"display":null,"left":null}},"8a2efb2cf0314c1e9eb61b70e75fd7e3":{"model_module":"@jupyter-widgets/controls","model_name":"DescriptionStyleModel","state":{"_view_name":"StyleView","_model_name":"DescriptionStyleModel","description_width":"","_view_module":"@jupyter-widgets/base","_model_module_version":"1.5.0","_view_count":null,"_view_module_version":"1.2.0","_model_module":"@jupyter-widgets/controls"}},"efa6b686ff5a4c609612fe0f95241633":{"model_module":"@jupyter-widgets/base","model_name":"LayoutModel","state":{"_view_name":"LayoutView","grid_template_rows":null,"right":null,"justify_content":null,"_view_module":"@jupyter-widgets/base","overflow":null,"_model_module_version":"1.2.0","_view_count":null,"flex_flow":null,"width":null,"min_width":null,"border":null,"align_items":null,"bottom":null,"_model_module":"@jupyter-widgets/base","top":null,"grid_column":null,"overflow_y":null,"overflow_x":null,"grid_auto_flow":null,"grid_area":null,"grid_template_columns":null,"flex":null,"_model_name":"LayoutModel","justify_items":null,"grid_row":null,"max_height":null,"align_content":null,"visibility":null,"align_self":null,"height":null,"min_height":null,"padding":null,"grid_auto_rows":null,"grid_gap":null,"max_width":null,"order":null,"_view_module_version":"1.2.0","grid_template_areas":null,"object_position":null,"object_fit":null,"grid_auto_columns":null,"margin":null,"display":null,"left":null}}}}},"cells":[{"cell_type":"markdown","metadata":{"id":"l4hucZoMjoKt"},"source":["# Data & Packages"]},{"cell_type":"code","metadata":{"id":"xvjqdsc6OrVG"},"source":["# embedding\n","!pip install -U sentence-transformers\n","# dimensionality reduction\n","!pip install umap-learn\n","# clustering \n","!pip install hdbscan"],"execution_count":null,"outputs":[]},{"cell_type":"code","metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"wORRsYt3OnmG","executionInfo":{"status":"ok","timestamp":1618701778151,"user_tz":300,"elapsed":273,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}},"outputId":"92059a2b-006f-42bc-90f9-b15d5b23c5f7"},"source":["import pandas as pd\n","import numpy as np\n","import matplotlib.pyplot as plt\n","import os\n","import nltk\n","nltk.download('punkt')\n","from nltk.tokenize import sent_tokenize\n","from sklearn.feature_extraction.text import CountVectorizer\n","from google.colab import drive\n","\n","\n","from sentence_transformers import SentenceTransformer\n","import umap\n","import hdbscan"],"execution_count":3,"outputs":[{"output_type":"stream","text":["[nltk_data] Downloading package punkt to /root/nltk_data...\n","[nltk_data]   Package punkt is already up-to-date!\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"S2Oi1E-KOhak"},"source":["drive.mount('/content/drive')"],"execution_count":null,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"qu8V17uuJxIc"},"source":["## Save Raw Text Data"]},{"cell_type":"code","metadata":{"id":"p4S1BcJcJYpN"},"source":["# collect all raw data from the drive\n","\n","input_path = \"/content/drive/My Drive/Rebuild my Professor/Ratings/\"\n","files = [file for file in os.listdir(input_path) if file.endswith(\"csv\")]\n","data = pd.DataFrame()\n","for file in files:\n","  df = pd.read_csv(input_path + file, engine=\"python\", index_col = 0)\n","  data = data.append(df)"],"execution_count":null,"outputs":[]},{"cell_type":"code","metadata":{"id":"sHIgy8WUSKFz"},"source":["# read the collected documents\n","\n","sentences = [sent for review in data[\"rComments\"] if type(review)==str and review for sent in sent_tokenize(review)]\n","print(\"Length of comments collected:\", len(data[data[\"rComments\"].notnull()]), \"length of sentences collected\", len(sentences))\n","\n","sentences_small = sentences[::10] # scale down to reduce computation\n","print(len(sentences_small))"],"execution_count":null,"outputs":[]},{"cell_type":"code","metadata":{"id":"GQKAZpv4qWFP"},"source":["# save the tokenized sentences collected from \n","\n","with open(\"/content/drive/My Drive/Rebuild my Professor/experiment_sentences_original.npy\", \"wb\") as f:\n","  np.save(f, sentences)"],"execution_count":null,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"QvVzB6KTJ4MN"},"source":["## Load Text Data "]},{"cell_type":"code","metadata":{"id":"myPpNieJqhFA","executionInfo":{"status":"ok","timestamp":1618701968232,"user_tz":300,"elapsed":12556,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}}},"source":["with open(\"/content/drive/My Drive/Rebuild my Professor/experiment_sentences_original.npy\", \"rb\") as f:\n","  sentences = np.load(f)"],"execution_count":5,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"nVxeeTNYji0k"},"source":["# Embedding\n"]},{"cell_type":"markdown","metadata":{"id":"ybhF2aLaLMrm"},"source":["## Save Embedding"]},{"cell_type":"code","metadata":{"id":"25bi6hbgZvAV"},"source":["# use Distilbert as it gives a nice balance between speed and performance\n","\n","model = SentenceTransformer('distilbert-base-nli-mean-tokens')"],"execution_count":null,"outputs":[]},{"cell_type":"code","metadata":{"id":"L7ZJ55oqiuR2"},"source":["# perform embedding on the sentences\n","\n","embeddings = model.encode(sentences_small, show_progress_bar=True)"],"execution_count":null,"outputs":[]},{"cell_type":"code","metadata":{"id":"6GAIFfArhhgT"},"source":["# save the embeddings locally for future runs\n","\n","with open(\"/content/drive/My Drive/Rebuild my Professor/experiment_embeddings_xue.npy\", \"wb\") as f:\n","  np.save(f, embeddings)"],"execution_count":null,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"IVIS5bX7LZvo"},"source":["## Load Sentence Embedding"]},{"cell_type":"code","metadata":{"id":"Vz3N9_9gj1Af","executionInfo":{"status":"ok","timestamp":1618702376182,"user_tz":300,"elapsed":23520,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}}},"source":["with open(\"/content/drive/My Drive/Rebuild my Professor/experiment_embeddings_xue.npy\", \"rb\") as f:\n","  embeddings = np.load(f)"],"execution_count":6,"outputs":[]},{"cell_type":"code","metadata":{"id":"vptjFNWw8F9w","executionInfo":{"status":"ok","timestamp":1618702599931,"user_tz":300,"elapsed":225,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}}},"source":["# furthur scale down to fit UMAP algorithm\n","embeddings = embeddings[::10]\n","sentences = sentences[::10]"],"execution_count":7,"outputs":[]},{"cell_type":"code","metadata":{"id":"IagYOV5XGJCk","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1618702601861,"user_tz":300,"elapsed":225,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}},"outputId":"05e8a975-332f-4e99-ba4e-6eb3b2b9482a"},"source":["print(\"The original length is\", len(embeddings))"],"execution_count":8,"outputs":[{"output_type":"stream","text":["The original length is 77353\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"IMCZvk3_7etp","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1618702720887,"user_tz":300,"elapsed":327,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}},"outputId":"f002c6da-d9e8-40c6-ef13-af1fc1f79873"},"source":["import string\n","\n","# data preprossing in sentences and embeddings\n","# remove sentences that only has punctuation\n","empty_count = 0\n","space_count = 0\n","punkt_count = 0\n","total_count = 0\n","flags = np.zeros(len(sentences), dtype=bool)\n","\n","for i in range(len(sentences)):\n","  cur_sentence = sentences[i]\n","  if not cur_sentence: \n","    empty_count += 1\n","  elif not cur_sentence.strip():\n","    space_count += 1\n","  elif cur_sentence[0] in string.punctuation:\n","    punkt_count += 1\n","  else:\n","    flags[i] = True\n","    continue\n","  \n","  total_count += 1\n","\n","print(\"Sentences that are empty:\", empty_count, \". Sentences that only has white space:\", space_count, \". Sentences that starts with punctuation:\", punkt_count)\n","print(\"Total count\", total_count, \"should be equal to the sum of the above\")"],"execution_count":11,"outputs":[{"output_type":"stream","text":["Sentences that are empty: 0 . Sentences that only has white space: 0 . Sentences that starts with punctuation: 903\n","Total count 903 should be equal to the sum of the above\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"fRLHIVcIHdD5","executionInfo":{"status":"ok","timestamp":1618702728045,"user_tz":300,"elapsed":420,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}}},"source":["# apply boolean indexing to sentences and embedding\n","sentences = sentences[flags]\n","embeddings = embeddings[flags]"],"execution_count":12,"outputs":[]},{"cell_type":"code","metadata":{"id":"myjEmCkOINEr","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1618702791505,"user_tz":300,"elapsed":207,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}},"outputId":"44f631af-0407-411e-8265-d0335e2ee62f"},"source":["# check the shape of both raw sentences and its embeddings\n","print(sentences.shape)\n","print(embeddings.shape)"],"execution_count":14,"outputs":[{"output_type":"stream","text":["(76450,)\n","(76450, 768)\n"],"name":"stdout"}]},{"cell_type":"markdown","metadata":{"id":"bdd_AAfXjhB8"},"source":["# Clustering"]},{"cell_type":"markdown","metadata":{"id":"24bJmNUTrGTe"},"source":["## UMAP\n","\n","Reference [Basic UMAP Parameters](https://umap-learn.readthedocs.io/en/latest/parameters.html) to tune the parameters.\n"," \n","`n_components` stands for the number of dimensionalities after reduction. If dimensionality is too low, there might be too much loss of information; whereas if it is too high, we might get poorer clustering results.`n_neighbors` controls the tradeoff between local structure versus global structure in the data. "]},{"cell_type":"code","metadata":{"id":"3ExQeCG2iyk5","executionInfo":{"status":"ok","timestamp":1618703636669,"user_tz":300,"elapsed":224277,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}}},"source":["# play around with these values to optimize for topic creation\n","umap_embeddings = umap.UMAP(n_neighbors=24, \n","                            n_components=15,\n","                            # min_dist=0.24, \n","                            metric='cosine').fit_transform(embeddings)"],"execution_count":15,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"A_hyyr9zrHej"},"source":["## HDBSCAN\n","\n","HDBSCAN is a density-based algorithm that works quite well with UMAP since UMAP maintains a lot of local structure even in lower-dimensional space. Besides, HDBSCAN does not force data points to clusters as it considers them outliers.\n","\n","Reference [Parameter Selection for HDBSCAN](https://hdbscan.readthedocs.io/en/latest/parameter_selection.html) to tune the parameters."]},{"cell_type":"code","metadata":{"id":"18dlx8_4rJQV","executionInfo":{"status":"ok","timestamp":1618703694041,"user_tz":300,"elapsed":18673,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}}},"source":["cluster = hdbscan.HDBSCAN(min_cluster_size= 24,\n","                          min_samples = 15,\n","                          # cluster_selection_epsilon = 0.2,\n","                          metric='euclidean',                      \n","                          cluster_selection_method='eom').fit(umap_embeddings)"],"execution_count":16,"outputs":[]},{"cell_type":"code","metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"9jR81tuDRs9A","executionInfo":{"status":"ok","timestamp":1618703694042,"user_tz":300,"elapsed":17619,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}},"outputId":"e8afdde3-6314-4e24-d9e5-1f75b27a2ad5"},"source":["print(\"Number of topics:\", cluster.labels_.max())"],"execution_count":17,"outputs":[{"output_type":"stream","text":["Number of topics: 291\n"],"name":"stdout"}]},{"cell_type":"markdown","metadata":{"id":"PujCj_mcuKgf"},"source":["# Visualization\n","\n","Although we can see it is difficult to visualize the individual clusters due to the number of topics generated (~300). However, we can see that even in 2-dimensional space, the algorithms keep some local structure."]},{"cell_type":"code","metadata":{"colab":{"base_uri":"https://localhost:8080/","height":595},"id":"roEdzs6ksB4i","executionInfo":{"status":"ok","timestamp":1618703806742,"user_tz":300,"elapsed":128304,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}},"outputId":"a1501ef3-8caf-4838-9242-5b439a19c6fe"},"source":["# To visualize the resulting clusters, reduce the dimensionality to 2 and visualize the outliers as grey points\n","\n","# Prepare data\n","umap_data = umap.UMAP(n_neighbors=15, n_components=2, min_dist=0.0, metric='cosine').fit_transform(embeddings)\n","result = pd.DataFrame(umap_data, columns=['x', 'y'])\n","result['labels'] = cluster.labels_\n","\n","# Visualize clusters\n","fig, ax = plt.subplots(figsize=(20, 10))\n","outliers = result.loc[result.labels == -1, :]\n","clustered = result.loc[result.labels != -1, :]\n","plt.scatter(outliers.x, outliers.y, color='#BDBDBD', s=0.05)\n","plt.scatter(clustered.x, clustered.y, c=clustered.labels, s=0.05, cmap='hsv_r')\n","plt.colorbar()"],"execution_count":18,"outputs":[{"output_type":"execute_result","data":{"text/plain":["<matplotlib.colorbar.Colorbar at 0x7f483530e510>"]},"metadata":{"tags":[]},"execution_count":18},{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 1440x720 with 2 Axes>"]},"metadata":{"tags":[],"needs_background":"light"}}]},{"cell_type":"markdown","metadata":{"id":"lw3i1xPfv2q8"},"source":["# Topic Creation"]},{"cell_type":"markdown","metadata":{"id":"Lj4ey1N0wwe2"},"source":["## c-TF-IDF"]},{"cell_type":"code","metadata":{"id":"RsU_JUMdv0mE","executionInfo":{"status":"ok","timestamp":1618703845792,"user_tz":300,"elapsed":394,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}}},"source":["docs_df = pd.DataFrame(sentences, columns=[\"Doc\"])\n","docs_df['Topic'] = cluster.labels_\n","docs_df['Doc_ID'] = range(len(docs_df))\n","docs_per_topic = docs_df.groupby(['Topic'], as_index = False).agg({'Doc': ' '.join})"],"execution_count":19,"outputs":[]},{"cell_type":"code","metadata":{"id":"Ly3o7vuXwrvT","executionInfo":{"status":"ok","timestamp":1618703847677,"user_tz":300,"elapsed":1685,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}}},"source":["def c_tf_idf(documents, m, ngram_range=(1, 1)):\n","    count = CountVectorizer(ngram_range=ngram_range, stop_words=\"english\").fit(documents)\n","    t = count.transform(documents).toarray()\n","    w = t.sum(axis=1)\n","    tf = np.divide(t.T, w)\n","    sum_t = t.sum(axis=0)\n","    idf = np.log(np.divide(m, sum_t)).reshape(-1, 1)\n","    tf_idf = np.multiply(tf, idf)\n","\n","    return tf_idf, count\n","  \n","\n","tf_idf, count = c_tf_idf(docs_per_topic.Doc.values, m=len(sentences[::10]))"],"execution_count":20,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"09rcEOdowzNP"},"source":["## Topic Representation\n"]},{"cell_type":"code","metadata":{"colab":{"base_uri":"https://localhost:8080/","height":514},"id":"cg3Thkjfwz0b","executionInfo":{"status":"ok","timestamp":1618703849141,"user_tz":300,"elapsed":394,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}},"outputId":"314b27c4-e783-48ca-8c95-29e1f573031d"},"source":["def extract_top_n_words_per_topic(tf_idf, count, docs_per_topic, n=20):\n","    words = count.get_feature_names()\n","    labels = list(docs_per_topic.Topic)\n","    tf_idf_transposed = tf_idf.T\n","    indices = tf_idf_transposed.argsort()[:, -n:]\n","    top_n_words = {label: [(words[j], tf_idf_transposed[i][j]) for j in indices[i]][::-1] for i, label in enumerate(labels)}\n","    return top_n_words\n","\n","def extract_topic_sizes(df):\n","    topic_sizes = (df.groupby(['Topic'])\n","                     .Doc\n","                     .count()\n","                     .reset_index()\n","                     .rename({\"Topic\": \"Topic\", \"Doc\": \"Size\"}, axis='columns')\n","                     .sort_values(\"Size\", ascending=False))\n","    return topic_sizes\n","\n","top_n_words = extract_top_n_words_per_topic(tf_idf, count, docs_per_topic, n=20)\n","topic_sizes = extract_topic_sizes(docs_df); topic_sizes.head(15)"],"execution_count":21,"outputs":[{"output_type":"execute_result","data":{"text/html":["<div>\n","<style scoped>\n","    .dataframe tbody tr th:only-of-type {\n","        vertical-align: middle;\n","    }\n","\n","    .dataframe tbody tr th {\n","        vertical-align: top;\n","    }\n","\n","    .dataframe thead th {\n","        text-align: right;\n","    }\n","</style>\n","<table border=\"1\" class=\"dataframe\">\n","  <thead>\n","    <tr style=\"text-align: right;\">\n","      <th></th>\n","      <th>Topic</th>\n","      <th>Size</th>\n","    </tr>\n","  </thead>\n","  <tbody>\n","    <tr>\n","      <th>0</th>\n","      <td>-1</td>\n","      <td>50410</td>\n","    </tr>\n","    <tr>\n","      <th>249</th>\n","      <td>248</td>\n","      <td>1640</td>\n","    </tr>\n","    <tr>\n","      <th>198</th>\n","      <td>197</td>\n","      <td>651</td>\n","    </tr>\n","    <tr>\n","      <th>89</th>\n","      <td>88</td>\n","      <td>619</td>\n","    </tr>\n","    <tr>\n","      <th>234</th>\n","      <td>233</td>\n","      <td>507</td>\n","    </tr>\n","    <tr>\n","      <th>207</th>\n","      <td>206</td>\n","      <td>488</td>\n","    </tr>\n","    <tr>\n","      <th>102</th>\n","      <td>101</td>\n","      <td>431</td>\n","    </tr>\n","    <tr>\n","      <th>91</th>\n","      <td>90</td>\n","      <td>406</td>\n","    </tr>\n","    <tr>\n","      <th>1</th>\n","      <td>0</td>\n","      <td>386</td>\n","    </tr>\n","    <tr>\n","      <th>270</th>\n","      <td>269</td>\n","      <td>370</td>\n","    </tr>\n","    <tr>\n","      <th>141</th>\n","      <td>140</td>\n","      <td>342</td>\n","    </tr>\n","    <tr>\n","      <th>292</th>\n","      <td>291</td>\n","      <td>336</td>\n","    </tr>\n","    <tr>\n","      <th>205</th>\n","      <td>204</td>\n","      <td>335</td>\n","    </tr>\n","    <tr>\n","      <th>290</th>\n","      <td>289</td>\n","      <td>327</td>\n","    </tr>\n","    <tr>\n","      <th>187</th>\n","      <td>186</td>\n","      <td>325</td>\n","    </tr>\n","  </tbody>\n","</table>\n","</div>"],"text/plain":["     Topic   Size\n","0       -1  50410\n","249    248   1640\n","198    197    651\n","89      88    619\n","234    233    507\n","207    206    488\n","102    101    431\n","91      90    406\n","1        0    386\n","270    269    370\n","141    140    342\n","292    291    336\n","205    204    335\n","290    289    327\n","187    186    325"]},"metadata":{"tags":[]},"execution_count":21}]},{"cell_type":"markdown","metadata":{"id":"xYUDgGOiRMQj"},"source":["## Reduce Cluster Number"]},{"cell_type":"code","metadata":{"id":"iqpybounnLac","colab":{"base_uri":"https://localhost:8080/","height":514},"executionInfo":{"status":"ok","timestamp":1618703934296,"user_tz":300,"elapsed":40287,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}},"outputId":"d695c81c-77ba-4856-ba2c-d5446dd565a8"},"source":["from sklearn.metrics.pairwise import cosine_similarity\n","\n","for i in range(24):\n","\n","    # Calculate cosine similarity\n","    similarities = cosine_similarity(tf_idf.T)\n","    np.fill_diagonal(similarities, 0)\n","\n","    # Extract label to merge into and from where\n","    topic_sizes = docs_df.groupby(['Topic']).count().sort_values(\"Doc\", ascending=False).reset_index()\n","    topic_to_merge = topic_sizes.iloc[-1].Topic\n","    topic_to_merge_into = np.argmax(similarities[topic_to_merge + 1]) - 1\n","\n","    # Adjust topics\n","    docs_df.loc[docs_df.Topic == topic_to_merge, \"Topic\"] = topic_to_merge_into\n","    old_topics = docs_df.sort_values(\"Topic\").Topic.unique()\n","    map_topics = {old_topic: index - 1 for index, old_topic in enumerate(old_topics)}\n","    docs_df.Topic = docs_df.Topic.map(map_topics)\n","    docs_per_topic = docs_df.groupby(['Topic'], as_index = False).agg({'Doc': ' '.join})\n","\n","    # Calculate new topic words\n","    m = len(sentences)\n","    tf_idf, count = c_tf_idf(docs_per_topic.Doc.values, m)\n","    top_n_words = extract_top_n_words_per_topic(tf_idf, count, docs_per_topic, n=20)\n","\n","topic_sizes = extract_topic_sizes(docs_df); topic_sizes.head(15)\n"],"execution_count":23,"outputs":[{"output_type":"execute_result","data":{"text/html":["<div>\n","<style scoped>\n","    .dataframe tbody tr th:only-of-type {\n","        vertical-align: middle;\n","    }\n","\n","    .dataframe tbody tr th {\n","        vertical-align: top;\n","    }\n","\n","    .dataframe thead th {\n","        text-align: right;\n","    }\n","</style>\n","<table border=\"1\" class=\"dataframe\">\n","  <thead>\n","    <tr style=\"text-align: right;\">\n","      <th></th>\n","      <th>Topic</th>\n","      <th>Size</th>\n","    </tr>\n","  </thead>\n","  <tbody>\n","    <tr>\n","      <th>0</th>\n","      <td>-1</td>\n","      <td>50459</td>\n","    </tr>\n","    <tr>\n","      <th>222</th>\n","      <td>221</td>\n","      <td>1640</td>\n","    </tr>\n","    <tr>\n","      <th>176</th>\n","      <td>175</td>\n","      <td>651</td>\n","    </tr>\n","    <tr>\n","      <th>81</th>\n","      <td>80</td>\n","      <td>644</td>\n","    </tr>\n","    <tr>\n","      <th>210</th>\n","      <td>209</td>\n","      <td>507</td>\n","    </tr>\n","    <tr>\n","      <th>184</th>\n","      <td>183</td>\n","      <td>488</td>\n","    </tr>\n","    <tr>\n","      <th>94</th>\n","      <td>93</td>\n","      <td>431</td>\n","    </tr>\n","    <tr>\n","      <th>83</th>\n","      <td>82</td>\n","      <td>406</td>\n","    </tr>\n","    <tr>\n","      <th>1</th>\n","      <td>0</td>\n","      <td>386</td>\n","    </tr>\n","    <tr>\n","      <th>240</th>\n","      <td>239</td>\n","      <td>370</td>\n","    </tr>\n","    <tr>\n","      <th>182</th>\n","      <td>181</td>\n","      <td>361</td>\n","    </tr>\n","    <tr>\n","      <th>127</th>\n","      <td>126</td>\n","      <td>342</td>\n","    </tr>\n","    <tr>\n","      <th>261</th>\n","      <td>260</td>\n","      <td>336</td>\n","    </tr>\n","    <tr>\n","      <th>259</th>\n","      <td>258</td>\n","      <td>327</td>\n","    </tr>\n","    <tr>\n","      <th>168</th>\n","      <td>167</td>\n","      <td>325</td>\n","    </tr>\n","  </tbody>\n","</table>\n","</div>"],"text/plain":["     Topic   Size\n","0       -1  50459\n","222    221   1640\n","176    175    651\n","81      80    644\n","210    209    507\n","184    183    488\n","94      93    431\n","83      82    406\n","1        0    386\n","240    239    370\n","182    181    361\n","127    126    342\n","261    260    336\n","259    258    327\n","168    167    325"]},"metadata":{"tags":[]},"execution_count":23}]},{"cell_type":"code","metadata":{"id":"iLeke84eo69q","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1618703963059,"user_tz":300,"elapsed":216,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}},"outputId":"954730c2-b9d1-41b3-de95-622b0244bd51"},"source":["print(\"Topic Number after reduction:\", topic_sizes.shape[0])"],"execution_count":24,"outputs":[{"output_type":"stream","text":["Topic Number after reduction: 262\n"],"name":"stdout"}]},{"cell_type":"markdown","metadata":{"id":"klT7ERBJ7U_W"},"source":["# Display Clusters for Topic Analysis"]},{"cell_type":"markdown","metadata":{"id":"1yG8uJO7hbLL"},"source":["## Print out the Clusters"]},{"cell_type":"code","metadata":{"id":"73xHZUgLrCcc","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1618704424859,"user_tz":300,"elapsed":231,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}},"outputId":"f7a7b8de-0e75-4181-9213-32172b3f4436"},"source":["from pprint import pprint\n","\n","index = topic_sizes['Topic'].iloc[1]\n","\n","pprint(top_n_words[index][:12])\n","print()\n","\n","pprint([sent for sent in sent_tokenize(docs_per_topic.Doc[index+1][:1200])], width=180)"],"execution_count":29,"outputs":[{"output_type":"stream","text":["[('final', 0.1201473873983707),\n"," ('week', 0.11782515010429576),\n"," ('grade', 0.0826900232734874),\n"," ('quizzes', 0.07986902243509132),\n"," ('weekly', 0.06912095899838884),\n"," ('10', 0.0670852583480535),\n"," ('exams', 0.058780231557535846),\n"," ('papers', 0.05850861634397788),\n"," ('paper', 0.05838745403317091),\n"," ('tests', 0.05407990374999552),\n"," ('midterm', 0.05254427113498678),\n"," ('20', 0.04960321053477523)]\n","\n","['5 easy quizzes (3 at home online) (4 questions each) and a final.',\n"," 'For the tests he chooses 5 random pieces from about 100.',\n"," 'Our second midterm average was a 28%.',\n"," 'The computer science department is week to begin with.',\n"," '3 exams with a cumulative final (all multiple choice).',\n"," 'Your grade is based on four papers, as far as the readings go you only need to read what goes with the essays you want to write.',\n"," \"The averages throughout the year so far have been 60's on everything.\",\n"," 'Three exams (lowest can be replaced by clicker score) and a final make up your grade.',\n"," 'Three 5-page papers, but he gives you study sheets.',\n"," 'There are only 3 grades (including the final) so you need to be good at multiple choice tests to do well.',\n"," 'Thursdays are essentially a group quiz that lasts the whole period.',\n"," 'Balaban also gave three extra credit chances this semester!',\n"," 'Final was 10 questions long.',\n"," \"3 of the 9 questions on the final were the exact 3 I didn't study on a list of 100 study questions.\",\n"," 'The class is also only graded on 3 papers, a midterm and a final.',\n"," 'Made one lecture async and we watched a movie for the other every week.',\n"," 'However the grade for Poli101 is entirely based on about 6 grades, three of which are grade']\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"jJNdOoD5OSvA","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1618704436081,"user_tz":300,"elapsed":219,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}},"outputId":"70f7569e-6274-4dcf-8c16-6d80dbc57f01"},"source":["index = topic_sizes['Topic'].iloc[2]\n","\n","pprint(top_n_words[index][:12])\n","print()\n","\n","pprint([sent for sent in sent_tokenize(docs_per_topic.Doc[index+1][:1200])], width=180)"],"execution_count":31,"outputs":[{"output_type":"stream","text":["[('students', 0.5096832830853644),\n"," ('cares', 0.4138028758043638),\n"," ('really', 0.13925662658345947),\n"," ('wants', 0.11107337995868823),\n"," ('genuinely', 0.10103383746893903),\n"," ('truly', 0.08286479359019852),\n"," ('student', 0.07841861751042477),\n"," ('succeed', 0.07405672507750563),\n"," ('help', 0.07060713481111071),\n"," ('willing', 0.04715996490375479),\n"," ('success', 0.043141582928783666),\n"," ('loves', 0.03603721124001564)]\n","\n","['Sokalski is engaging and is invested in student success.',\n"," 'Prof. Cahen is a very knowledgeable guy and sincerely cares about how his students do.',\n"," 'He really cares about his students and loves to get involved in extracurricular projects.',\n"," \"His door is always open to students and he's very willing to help with anything.\",\n"," \"One of the few genuinely good people I know ... actually cares about teaching the subject and the people he's teaching.\",\n"," 'Bulent was enthusiastic, very helpful and clearly cares about his students.',\n"," 'Also, he really cares about his students!',\n"," 'He really cares about his students and wants them to do well.',\n"," \"He really cares about his students' success and he's available frequently outside of class.\",\n"," 'Professor Curley is very nice and respectful of his students.',\n"," 'Andrew consistently goes above and beyond the call of duty to helps students Helps students engage and understand.',\n"," 'He really cares about his students and if you ever have questions he is very approachable.',\n"," 'He really cares about engaging with his students!',\n"," 'He genuinely cares about each student doing well and making the material understandable.',\n"," 'Participation matters and he appreciates the students who engage.',\n"," 'He does truly care a']\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"7e8yIRKmxLzv","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1618704446499,"user_tz":300,"elapsed":260,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}},"outputId":"4b1a29c1-5250-404e-f8d0-f31aa146b8c4"},"source":["index = topic_sizes['Topic'].iloc[3]\n","\n","pprint(top_n_words[index][:12])\n","print()\n","\n","pprint([sent for sent in sent_tokenize(docs_per_topic.Doc[index+1][:1200])], width=180)"],"execution_count":32,"outputs":[{"output_type":"stream","text":["[('best', 0.6620235395995314),\n"," ('professors', 0.35496871863090235),\n"," ('professor', 0.3066532985401281),\n"," ('ve', 0.29946462994849216),\n"," ('far', 0.15620768243588268),\n"," ('dr', 0.0741120442177326),\n"," ('probably', 0.0567959629887183),\n"," ('greatest', 0.05507231490347427),\n"," ('hands', 0.048693659907009086),\n"," ('prof', 0.04652135019544428),\n"," ('years', 0.04502697832160136),\n"," ('teachers', 0.04301078199026585)]\n","\n","[\"Deb is the BEST professor i've had in college.\",\n"," 'I honestly can say that Professor Mackey is the best professor I have ever had.',\n"," \"Easily one of the best professors I've had.\",\n"," 'best professor ever One of the best professors I have ever had.',\n"," \"One of the greatest professors I've ever had.\",\n"," 'Professor Walsh is absolutely one of the best professors at UVA.',\n"," \"Professor Boettiger is one of the best professors I've had so far at UNC.\",\n"," 'Overall, probably the best choice for biochem at UNC.',\n"," 'She is probably one of the best professors for ENGL105.',\n"," 'He is absolutely one of the best professors at UNC and one of the best professors in the United States.',\n"," 'Brian Hogan is the best professor in all of the chem department.',\n"," 'Prof. Kasson is the best teacher I had a Carolina.',\n"," \"Laurie Paul is one of the best professors I've ever had.\",\n"," 'Absolutely the best professor I have had so far at UNC.',\n"," \"Professor Warner is easily the best prof I've had at UNC.\",\n"," 'He is the most knowledgable Italian professor I have ever had the honor of meeting.',\n"," 'While most Arabic profs at NYU are worth your time, Ali should be at the top of your list.',\n"," \"Best professor Andre is the best professor I've had so far.\",\n"," 'He is the best professor I ever had at NYU!']\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"3aylg7KAxdhy","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1618704458928,"user_tz":300,"elapsed":209,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}},"outputId":"c6b8b300-ca9b-45c2-e618-b2a97b1b8306"},"source":["index = topic_sizes['Topic'].iloc[4]\n","\n","pprint(top_n_words[index][:12])\n","print()\n","\n","pprint([sent for sent in sent_tokenize(docs_per_topic.Doc[index+1][:1200])], width=180)"],"execution_count":34,"outputs":[{"output_type":"stream","text":["[('read', 0.14908319290045433),\n"," ('book', 0.129402928240613),\n"," ('class', 0.11453582194368048),\n"," ('notes', 0.09365354339555523),\n"," ('study', 0.09133202455353284),\n"," ('textbook', 0.08295351625494601),\n"," ('exams', 0.08014249397011583),\n"," ('homework', 0.07611833161313411),\n"," ('lecture', 0.06702113408908439),\n"," ('sure', 0.06314465107323121),\n"," ('exam', 0.05964142135055464),\n"," ('tests', 0.05737864933898007)]\n","\n","['If you complete the hw and understand it, the exams are simply re-worded hw probs.',\n"," 'Just make sure you do the reading and go to class (a lot of the test questions come from lecture.)',\n"," 'Tests, rec sec, and papers are your grade.',\n"," 'Just go to class and do the readings.',\n"," \"Go to class and you'll know the information!\",\n"," 'You just gotta show up and listen to the book.',\n"," 'That being said, get ready to read if you take this class.',\n"," 'Discussions =professor trying to relate the topics that we are discussing in class.',\n"," 'She goes straight by the book.',\n"," 'He basically recites the readings in lecture.',\n"," 'Go to lecture if you want to see examples done in class.',\n"," 'Just read the textbook.',\n"," \"Take his class if you're into reading and anti-climatic jokes.\",\n"," 'If you want to get anything out of his lectures come to class prepared.',\n"," 'TEXTBOOK  LIFELINE.',\n"," 'Do the readings before class.',\n"," 'Watch out for the exams though.',\n"," 'If you want to memorize her book and repeat it verbatim on exams, take this class.',\n"," 'just study his study guide and youre golden!',\n"," 'you should read textbook.',\n"," 'Do your webwork, go to class, and show up to section.',\n"," 'Come to class prepared with things to say.',\n"," 'Just be sure to do the homework, remember the quizzes, and go to class.',\n"," 'hw i']\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"FzxTA6Ouxduf","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1618704469472,"user_tz":300,"elapsed":321,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}},"outputId":"57d04c34-3cb8-48aa-ba1a-67175323474e"},"source":["index = topic_sizes['Topic'].iloc[5]\n","\n","pprint(top_n_words[index][:12])\n","print()\n","\n","pprint([sent for sent in sent_tokenize(docs_per_topic.Doc[index+1][:1200])], width=180)"],"execution_count":35,"outputs":[{"output_type":"stream","text":["[('class', 0.34220168560948927),\n"," ('fun', 0.3176977044850838),\n"," ('interesting', 0.2707969370599525),\n"," ('great', 0.11941313398419366),\n"," ('makes', 0.10132210042456181),\n"," ('good', 0.09083776501146486),\n"," ('overall', 0.08170898315779579),\n"," ('funny', 0.07120109953094952),\n"," ('amazing', 0.06698718178761671),\n"," ('awesome', 0.06243739273682795),\n"," ('enjoyable', 0.06056847612724808),\n"," ('really', 0.05630854546910087)]\n","\n","['Very fun and chill class The in-class composition projects are a lot of fun as well.',\n"," 'A lot of fun, a really interesting class, taught by a man who knows his stuff and can present it clearly.',\n"," 'The class is interesting and he is really funny.',\n"," 'This class was awesome!',\n"," 'his sense of humor makes class fun.',\n"," 'This is a great class.',\n"," \"Aside from grades, hes very entertaining cultural events and time periods you'll enjoy this class and he makes it very interesting.\",\n"," 'Super interesting class.',\n"," \"A good class to take 'for fun'.\",\n"," 'a very interesting class!',\n"," 'such a great guy that makes an 8am class so enjoyable.',\n"," 'He makes class fun and interesting.',\n"," 'Awesome class.',\n"," 'and honestly he made learning about the brain fun and interesting.',\n"," 'He made the class enjoyable and insightful.',\n"," 'Such an amazing help with chem!',\n"," 'This was an amazing class.',\n"," 'a truly interesting class that proved to be way more fun than I anticipated.',\n"," 'Class is fun and interesting.',\n"," 'This class is so much fun.',\n"," 'He was pretty funny and wanted to make class fun.',\n"," 'His class was awesome!',\n"," 'It was a fun and interesting class.',\n"," 'Awesome class.',\n"," \"Class is quite engaging though; he's pretty funny.\",\n"," 'This conversation class is amazing.',\n"," \"If you're interested in the subject, it s\"]\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"uW_ycbgRxd6V","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1618704474417,"user_tz":300,"elapsed":212,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}},"outputId":"f85a745a-43ea-485f-ca3a-184137d267cc"},"source":["index = topic_sizes['Topic'].iloc[6]\n","\n","pprint(top_n_words[index][:12])\n","print()\n","\n","pprint([sent for sent in sent_tokenize(docs_per_topic.Doc[index+1][:1200])], width=180)"],"execution_count":36,"outputs":[{"output_type":"stream","text":["[('worst', 0.99178356545613),\n"," ('professor', 0.28670081013231663),\n"," ('ve', 0.24772412850887374),\n"," ('teacher', 0.19602981248322224),\n"," ('far', 0.12042223155446584),\n"," ('professors', 0.09355702983837344),\n"," ('taken', 0.0772970752815421),\n"," ('ive', 0.05998407770108738),\n"," ('teachers', 0.05594750973360668),\n"," ('seen', 0.050936952265377686),\n"," ('university', 0.049851074775293824),\n"," ('life', 0.046393949865602906)]\n","\n","[\"Chang is the worst professor I've had at UVA!\",\n"," 'Without question, this is the worst class I have taken in my entire academic career.',\n"," 'WORST professor ever.',\n"," 'Worst prof in my Carolina career Best prof for the worst class.',\n"," \"Buck Goldstein's Econ 125 is one of the worst classes I have ever taken.\",\n"," \"One of the worst professors I've had at Carolina (I'm a Senior).\",\n"," \"Professor Ji is absolutely the worst professor I've had at UNC.\",\n"," \"Worst professor I've ever had in so many ways.\",\n"," \"Besides David Jones, the worst professor I've had at UNC.\",\n"," 'Professor Howe was the worst professor I have ever had at Wake.',\n"," 'He is definitely one of the worst professors I have ever taken, and I am comparing him to high school teachers and their assistants.',\n"," 'THE worst pro at NYU!',\n"," \"One of the worst teacher's I've ever had at NYU.\",\n"," 'Ross Geller is the worst professor I have ever had the displeasure of meeting.',\n"," 'He is the rudest professor I have ever encountered.',\n"," \"Maybe the worst Gallatin professor I've had.\",\n"," \"He is the absolute worst and most lazy professor I've ever had at NYU.\",\n"," \"Worse econ professor I've had at NYU.\",\n"," \"The buffer lab was the worst thing I've seen/done in my life.\",\n"," 'By far worst teacher Ive had.',\n"," \"Worst professor I've ever had a\"]\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"hNmHol-sxnkm","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1618704479729,"user_tz":300,"elapsed":206,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}},"outputId":"d6e70991-bb96-4bb1-9188-517ee628906f"},"source":["index = topic_sizes['Topic'].iloc[7]\n","\n","pprint(top_n_words[index][:12])\n","print()\n","\n","pprint([sent for sent in sent_tokenize(docs_per_topic.Doc[index+1][:1200])], width=180)"],"execution_count":37,"outputs":[{"output_type":"stream","text":["[('students', 0.5030536100176212),\n"," ('cares', 0.39159698697931655),\n"," ('really', 0.15649314760360009),\n"," ('wants', 0.11194656685320473),\n"," ('genuinely', 0.09825518420080627),\n"," ('truly', 0.08514729110756189),\n"," ('help', 0.06748816380814825),\n"," ('loves', 0.06541940250309423),\n"," ('student', 0.045757159589002804),\n"," ('passionate', 0.04471569677949641),\n"," ('success', 0.04086061356835146),\n"," ('succeed', 0.03762677648900657)]\n","\n","[\"She's very helpful and really cares how about your well-being in the class.\",\n"," 'She will do anything to help her students and is so clearly interested in and good at what she does!',\n"," \"She is very receptive to students' ideas and encourages independent research.\",\n"," \"She's hilarious and really cares about her students.\",\n"," 'Dr. Current genuinely cares about her students and wants them to succeed.',\n"," 'She is really caring for her students and will work with them fi a problem arises.',\n"," \"She's very nice and seems to care about the students.\",\n"," 'Dr. Haskin really seems to care about how her students do.',\n"," 'She encourages students to participate and kindly corrects you if there are mistakes.',\n"," 'S. Kim really cares about her students, and can be funny at times.',\n"," 'Expects her students to know all the intricacies of music like she does.',\n"," 'We talk often and I truly feel that she respects me not only as a student but as a person as well.',\n"," 'She is super sweet and genuinely cares about her students.',\n"," 'Although she does really care about her students, I would take 230 with someone else.',\n"," \"She really cares about the subject, and more importantly, the student's growth and progress.\",\n"," 'Prof.Davies is very willing to offer help to any of her stu']\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"OUMCAdO7xxJN","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1618704484147,"user_tz":300,"elapsed":224,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}},"outputId":"2405ed85-6c03-43d2-cc50-19e31037eafd"},"source":["index = topic_sizes['Topic'].iloc[8]\n","\n","pprint(top_n_words[index][:12])\n","print()\n","\n","pprint([sent for sent in sent_tokenize(docs_per_topic.Doc[index+1][:1200])], width=180)"],"execution_count":38,"outputs":[{"output_type":"stream","text":["[('comments', 4.893088811565543),\n"," ('projects', 0.014336697393736629),\n"," ('über', 0.0),\n"," ('examine', 0.0),\n"," ('exaggerating', 0.0),\n"," ('exaggeration', 0.0),\n"," ('exam', 0.0),\n"," ('exam2', 0.0),\n"," ('examens', 0.0),\n"," ('examination', 0.0),\n"," ('examined', 0.0),\n"," ('exaggerate', 0.0)]\n","\n","['No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No '\n"," 'Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments '\n"," 'No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No '\n"," 'Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments '\n"," 'No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No '\n"," 'Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments '\n"," 'No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments No Comments']\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"eTITPQ4UxyrD","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1618704492523,"user_tz":300,"elapsed":217,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}},"outputId":"5d75454d-5f6f-414b-b8e0-7f000f90121c"},"source":["index = topic_sizes['Topic'].iloc[9]\n","\n","pprint(top_n_words[index][:12])\n","print()\n","\n","pprint([sent for sent in sent_tokenize(docs_per_topic.Doc[index+1][:1200])], width=180)"],"execution_count":39,"outputs":[{"output_type":"stream","text":["[('good', 0.16604917474742525),\n"," ('disorganized', 0.14232992310135323),\n"," ('waste', 0.12611686152211),\n"," ('helpful', 0.10135757372406076),\n"," ('time', 0.0838200489823093),\n"," ('teacher', 0.06848266819711823),\n"," ('like', 0.06466114653216502),\n"," ('did', 0.06014116937626045),\n"," ('professor', 0.05750722726558777),\n"," ('lacks', 0.05069887724385696),\n"," ('unclear', 0.047693569495270316),\n"," ('didn', 0.04697410877127659)]\n","\n","['Not a fun class to be in.',\n"," \"Idk, she didn't make it interesting enough to keep my attention.\",\n"," 'He does not explain things well at all.',\n"," 'He is not good at teaching.',\n"," 'Pop quizzes were not good either.',\n"," 'Nobody did well on it.',\n"," 'Unfortunately  testsquizzes do not give partial credit.',\n"," 'Not a great professor.',\n"," 'Jed is not a great professor.',\n"," 'So unsatisfied.',\n"," 'Waste of time.',\n"," \"I don't know why her ratings are so low.\",\n"," \"Her advice and feedback wasn't helpful at all.\",\n"," 'She lacks professionalism and a passion for teaching.',\n"," \"First of all, she's NOT a good lecturer.\",\n"," 'Has a very scattered train of thought.',\n"," 'disliked working with her on such a close level.',\n"," \"Not too great That's just unprofessional.\",\n"," 'Very unapproachable and discouraging.',\n"," 'Section felt pointless.',\n"," 'Material and PPTs disorganized.',\n"," 'The class is a waste of time.',\n"," 'Ken was not so helpful for me as I hoped.',\n"," 'Reading was less interesting than I hoped.',\n"," \"He's not very helpful when it comes to answering your questions.\",\n"," 'She is not a good lecturer.',\n"," \"I don't know why she has bad ratings!\",\n"," \"I can't even say that they're hard, they're extremely random as a matter of fact!!!\",\n"," 'not good at all Section was kind of a waste of time.',\n"," 'Kim is interesting... Not the best professor...',\n"," 'Very c']\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"kHl5iD45x2tC","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1618704509918,"user_tz":300,"elapsed":261,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}},"outputId":"8b594178-354b-46b8-ae1e-1246d38356b1"},"source":["index = topic_sizes['Topic'].iloc[10]\n","\n","pprint(top_n_words[index][:12])\n","print()\n","\n","pprint([sent for sent in sent_tokenize(docs_per_topic.Doc[index+1][:1200])], width=180)"],"execution_count":40,"outputs":[{"output_type":"stream","text":["[('lectures', 0.46219646221108246),\n"," ('interesting', 0.2580566058329122),\n"," ('funny', 0.11460502316543943),\n"," ('engaging', 0.09991919060387647),\n"," ('great', 0.08827630125831078),\n"," ('fun', 0.07847033791295085),\n"," ('entertaining', 0.07212093217625516),\n"," ('lecture', 0.056560111550603843),\n"," ('makes', 0.05643672815231799),\n"," ('really', 0.05271896808409707),\n"," ('good', 0.05037661450540317),\n"," ('amazing', 0.04167046065560865)]\n","\n","[\"He also has a good sense of humor and knows when students aren't understanding his lecture.\",\n"," 'extremely interesting subject matter, interesting lectures.',\n"," 'The lectures were interesting and useful to attend.',\n"," \"His lectures are very interesting and he's a funny guy.\",\n"," 'Big Data &amp; Media Analysis was very interesting and super fun.',\n"," 'Lecture was well structured and interesting.',\n"," 'He does a great job of making you appreciate public policy.',\n"," 'Good lectures too.',\n"," 'His lectures are funny with personal anecdotes.',\n"," 'Gives both great lectures and feedback.',\n"," 'He made  lectures actually fascinating to attend.',\n"," \"He's good at teaching and explaining the themes behind the readings.\",\n"," 'Archeology should be something you are interested in but the lectures are fun, engaging, & full of witticisms.',\n"," 'Great time good material amazing lectures.',\n"," 'The rather ridiculous jokes made the lectures a lot more entertaining too!',\n"," 'Recorded lectures are great.',\n"," 'Very interesting lectures.',\n"," \"Lectures worth attending if u are not sure about anything, as he's super helpful &amp; amazing!\",\n"," 'Very knowledgeable and makes lectures interesting.',\n"," 'His lectures are very engaging and super interesting.',\n"," 'A lot of reading but his lectures are straightforward, pa']\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"-fz4Il_Lx3fz","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1618704515644,"user_tz":300,"elapsed":213,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}},"outputId":"0c3ff75a-3306-43b4-ef5e-a8634755ed1b"},"source":["index = topic_sizes['Topic'].iloc[11]\n","\n","pprint(top_n_words[index][:12])\n","print()\n","\n","pprint([sent for sent in sent_tokenize(docs_per_topic.Doc[index+1][:1200])], width=180)"],"execution_count":41,"outputs":[{"output_type":"stream","text":["[('bad', 0.4895968332591189),\n"," ('hard', 0.2610852666246978),\n"," ('isn', 0.1853086986514126),\n"," ('difficult', 0.1308889781882525),\n"," ('accent', 0.11335869112440544),\n"," ('aren', 0.10812048453364745),\n"," ('class', 0.10286387314722065),\n"," ('tests', 0.08666219290957876),\n"," ('wasn', 0.08219805532618266),\n"," ('overall', 0.07444688460390331),\n"," ('grader', 0.05210897289466212),\n"," ('really', 0.04575760104404362)]\n","\n","['A B is not hard.',\n"," \"I really didn't think Ramirez was that bad.\",\n"," 'Tests are not too difficult if you put in the time studying.',\n"," 'Not a hard class.',\n"," 'the subject isnt too hard to understand.',\n"," 'The class is good for a core requirement otherwise, nah.',\n"," \"She's not that bad.\",\n"," \"Prof's accent really isn't that bad.\",\n"," 'above average.',\n"," 'Not an impossible class!',\n"," \"Good grader, the work load wasn't too bad or heavy.\",\n"," 'Exams are not too hard.',\n"," 'Other than that, not a bad class.',\n"," 'Not bad to look at either!',\n"," \"Tests aren't too bad if you study.\",\n"," \"Her tests aren't very difficult.\",\n"," 'The work load was not excessive.',\n"," \"He's not bad to look at either!\",\n"," 'A fair amount of hw but not too difficult.',\n"," \"There wasn't many practice problems for the class.\",\n"," 'Class was not too difficult.',\n"," 'There is no curve in this class though.',\n"," \"He's not as bad as people make him out to be.\",\n"," 'The course work was not lengthy once you got into it.',\n"," 'The class difficulty is average.',\n"," 'Not as easy as I thought it would be but definitely not hard.',\n"," \"She's not that bad!\",\n"," 'The only problem is his accent.',\n"," \"It can get pretty tedious, but it's definitely an easy A.\",\n"," 'Not nearly as hard as people make it out to be.',\n"," \"but overall, it's not too difficult.\",\n"," \"Overall the class isn't bad.\",\n"," 'Not a bad pro']\n"],"name":"stdout"}]},{"cell_type":"markdown","metadata":{"id":"HIIfx6whhYa2"},"source":["## Saving Cluster"]},{"cell_type":"code","metadata":{"id":"mK07a0yiVlZ8"},"source":["# save the clusters \n","for index in range(1, 51):\n","  address = \"/content/drive/My Drive/Rebuild my Professor/Clusters_xue_Mar7/\" + str(index) + \".txt\"\n","  doc_index = topic_sizes['Topic'].iloc[index]\n","\n","  with open(address, \"w\") as f:\n","    f.write(docs_per_topic.Doc[doc_index+1])"],"execution_count":null,"outputs":[]},{"cell_type":"code","metadata":{"id":"mOebezvXk11K"},"source":["# try reading the clusters to make sure they are save in the correct format\n","for index in range(1, 51):\n","  address = \"/content/drive/My Drive/Rebuild my Professor/Clusters_xue_Mar7/\" + str(index) + \".txt\"\n","  doc_index = topic_sizes['Topic'].iloc[index]\n","  with open(address, \"r\") as f:\n","    print(f.read())"],"execution_count":null,"outputs":[]},{"cell_type":"code","metadata":{"id":"ySyhfSi_DGsS"},"source":["# save txt file above but we want csv file for readability\n","\n","# read txt file of each cluster and load into csv\n","for index in range(1, 51):\n","  address = \"/content/drive/My Drive/Rebuild my Professor/Clusters_xue_Mar7/\" + str(index) + \".txt\"\n","  with open(address, \"r\") as f:\n","    # tokenize\n","    sentences = sent_tokenize(f.read())\n","    df = pd.DataFrame(sentences)\n","    save_address = \"/content/drive/My Drive/Rebuild my Professor/Clusters_xue_Mar7/\" + str(index) + \".csv\"\n","    df.to_csv(save_address)"],"execution_count":null,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"VLUwWbjF63V6"},"source":["# Clustering on Selected Sentences\n","\n","- Participation\n","- Workload\n","\n","Since we don't find \"**participation**\" and \"**workload**\" clusters when we run the clustering algorithm above, we want to re-cluster some chosen sentences based on the two categories. We first select some sentences based on their words composition and store them in the JSON files. Then we perform the clustering algorithm again, hoping to gain more insights into the data regarding the two categories. \n","\n"]},{"cell_type":"markdown","metadata":{"id":"2T4EbU6Ka0pr"},"source":["### Workload"]},{"cell_type":"code","metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"ZEDF0hHD71yG","executionInfo":{"status":"ok","timestamp":1618705941946,"user_tz":300,"elapsed":838,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}},"outputId":"6474cf61-9f6d-4138-8fff-a32e11f92c89"},"source":["# read data\n","import json\n","\n","with open(\"/content/drive/My Drive/Rebuild my Professor/Workload.json\") as f:\n","  data = json.load(f)\n","  print(\"We read\",len(data), \"sentences regarding workload.\")"],"execution_count":42,"outputs":[{"output_type":"stream","text":["We read 113261 sentences regarding workload.\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"VCk16PXr-rz-","colab":{"base_uri":"https://localhost:8080/","height":66,"referenced_widgets":["4b0b89039f8840a1a8408945ca005475","ba6bac176442433a9e40ad20a0a4e081","ad7a571381454cc6bcf459aff99913c4","07e1a86a3b3c47b4a7991cbe24253c83","c8403bd4cc95435dbe6e247461772b38","b870e81c04f242f581fe33ae5708a8b3","6ecd6eacb06e4889bbfb7c947f584816","8307eaff07bc43c1b1fb033833599da8"]},"executionInfo":{"status":"ok","timestamp":1618705983726,"user_tz":300,"elapsed":19032,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}},"outputId":"d77a77ce-1c20-4785-8a17-25130569153f"},"source":["# import embedding model\n","model = SentenceTransformer('distilbert-base-nli-mean-tokens')"],"execution_count":46,"outputs":[{"output_type":"display_data","data":{"application/vnd.jupyter.widget-view+json":{"model_id":"4b0b89039f8840a1a8408945ca005475","version_minor":0,"version_major":2},"text/plain":["HBox(children=(FloatProgress(value=0.0, max=244733649.0), HTML(value='')))"]},"metadata":{"tags":[]}},{"output_type":"stream","text":["\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"pXKnLYVO8FpK","executionInfo":{"status":"ok","timestamp":1618705987783,"user_tz":300,"elapsed":219,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Ght_rN3bEDaFrWBGpcGJgaWkZz1zAHBA0_f6XWfFA=s64","userId":"16637980194371898177"}}},"source":["# function for c_tf_idf calculation\n","def c_tf_idf(documents, m, ngram_range=(1, 1)):\n","    count = CountVectorizer(ngram_range=ngram_range, stop_words=\"english\").fit(documents)\n","    t = count.transform(documents).toarray()\n","    w = t.sum(axis=1)\n","    tf = np.divide(t.T, w)\n","    sum_t = t.sum(axis=0)\n","    idf = np.log(np.divide(m, sum_t)).reshape(-1, 1)\n","    tf_idf = np.multiply(tf, idf)\n","\n","    return tf_idf, count\n","  \n","\n","# functions for topic representation\n","def extract_top_n_words_per_topic(tf_idf, count, docs_per_topic, n=20):\n","    words = count.get_feature_names()\n","    labels = list(docs_per_topic.Topic)\n","    tf_idf_transposed = tf_idf.T\n","    indices = tf_idf_transposed.argsort()[:, -n:]\n","    top_n_words = {label: [(words[j], tf_idf_transposed[i][j]) for j in indices[i]][::-1] for i, label in enumerate(labels)}\n","    return top_n_words\n","\n","\n","def extract_topic_sizes(df):\n","    topic_sizes = (df.groupby(['Topic'])\n","                     .Doc\n","                     .count()\n","                     .reset_index()\n","                     .rename({\"Topic\": \"Topic\", \"Doc\": \"Size\"}, axis='columns')\n","                     .sort_values(\"Size\", ascending=False))\n","    return topic_sizes\n","\n","\n","# function header for clustering based on input sentences\n","def topic_analysis(sentences):\n","    # embedding\n","    embeddings = model.encode(sentences, show_progress_bar=True)\n","\n","    # umap\n","    umap_embeddings = umap.UMAP(n_neighbors=24, \n","                            n_components=15,\n","                            # min_dist=0.24, \n","                            metric='cosine').fit_transform(embeddings)\n","    # hdbscan\n","    cluster = hdbscan.HDBSCAN(min_cluster_size= 24,\n","                          min_samples = 15,\n","                          # cluster_selection_epsilon = 0.2,\n","                          metric='euclidean',                      \n","                          cluster_selection_method='eom').fit(umap_embeddings)\n","                          \n","    # topic creation\n","    docs_df = pd.DataFrame(sentences, columns=[\"Doc\"])\n","    docs_df['Topic'] = cluster.labels_\n","    docs_df['Doc_ID'] = range(len(docs_df))\n","    docs_per_topic = docs_df.groupby(['Topic'], as_index = False).agg({'Doc': ' '.join})\n","\n","    tf_idf, count = c_tf_idf(docs_per_topic.Doc.values, m=len(sentences))\n","\n","    top_n_words = extract_top_n_words_per_topic(tf_idf, count, docs_per_topic, n=20)\n","    topic_sizes = extract_topic_sizes(docs_df); print(topic_sizes.head(15))\n","\n","    return docs_per_topic, top_n_words, topic_sizes"],"execution_count":47,"outputs":[]},{"cell_type":"code","metadata":{"id":"NBdMg0pG9usu"},"source":["docs_per_topic, top_n_words, topic_sizes = topic_analysis(data)"],"execution_count":null,"outputs":[]},{"cell_type":"code","metadata":{"id":"j5v3vb4TAJ_l"},"source":["# analyze clusters and choose one that is most relatable to workload\n","\n","from pprint import pprint\n","\n","index = topic_sizes['Topic'].iloc[6]\n","\n","pprint(top_n_words[index][:12])\n","print()\n","\n","pprint([sent for sent in sent_tokenize(docs_per_topic.Doc[index+1][:1200])])"],"execution_count":null,"outputs":[]},{"cell_type":"code","metadata":{"id":"sC_uTdfb_cqH"},"source":["# save #6 workload cluster above\n","\n","address = \"/content/drive/My Drive/Rebuild my Professor/Clusters_xue_Mar7/\" + \"workload\" + \".csv\"\n","doc_index = topic_sizes['Topic'].iloc[5]\n","sentences = sent_tokenize(docs_per_topic.Doc[doc_index+1])\n","\n","with open(address, \"w\") as f:\n","    df = pd.DataFrame(sentences)\n","    df.to_csv(address)"],"execution_count":null,"outputs":[]},{"cell_type":"markdown","metadata":{"id":"8PBeEW6wavv4"},"source":["### Participation"]},{"cell_type":"code","metadata":{"colab":{"base_uri":"https://localhost:8080/","height":389,"referenced_widgets":["ee2b5c1f3a1e4144827d553bfbecfb50","2ec2512f308f4c1aae3441191c8b220f","e5973bf0174c423daf5026098e6393c0","8f670ca1fa264a5884bef98b52ad52b6","7d889b10ac104d73abbc4349e9d350e3","65f28be5c5ab48c38b00b94726093434","8a2efb2cf0314c1e9eb61b70e75fd7e3","efa6b686ff5a4c609612fe0f95241633"]},"id":"P_rlqB3_jEpF","executionInfo":{"status":"ok","timestamp":1616028540521,"user_tz":300,"elapsed":4280801,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Gg3bNTF3nMtSb3LYsQrTsDM3WE_acmXU6ux-k4uKw=s64","userId":"09242844760289944401"}},"outputId":"19fb1c48-b929-4d35-8dc9-61c1f4a5365f"},"source":["with open(\"/content/drive/My Drive/Rebuild my Professor/Participation.json\") as f:\n","  data_par = json.load(f)\n","  print(\"We read\",len(data_par), \"sentences regarding participation.\")\n","\n","docs_per_topic, top_n_words, topic_sizes = topic_analysis(data_par)"],"execution_count":null,"outputs":[{"output_type":"stream","text":["We read 122597 sentences regarding participation.\n"],"name":"stdout"},{"output_type":"display_data","data":{"application/vnd.jupyter.widget-view+json":{"model_id":"ee2b5c1f3a1e4144827d553bfbecfb50","version_minor":0,"version_major":2},"text/plain":["HBox(children=(FloatProgress(value=0.0, description='Batches', max=3832.0, style=ProgressStyle(description_wid…"]},"metadata":{"tags":[]}},{"output_type":"stream","text":["\n"],"name":"stdout"},{"output_type":"stream","text":["/usr/local/lib/python3.7/dist-packages/ipykernel_launcher.py:6: RuntimeWarning: invalid value encountered in true_divide\n","  \n"],"name":"stderr"},{"output_type":"stream","text":["     Topic   Size\n","0       -1  81185\n","227    226   2256\n","91      90   1989\n","368    367   1224\n","61      60   1195\n","327    326   1001\n","47      46    935\n","333    332    841\n","153    152    760\n","89      88    753\n","259    258    653\n","1        0    641\n","351    350    534\n","188    187    527\n","236    235    514\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"lOp2ljAd2NjW","executionInfo":{"status":"ok","timestamp":1616044988867,"user_tz":300,"elapsed":461,"user":{"displayName":"Xue Zou","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14Gg3bNTF3nMtSb3LYsQrTsDM3WE_acmXU6ux-k4uKw=s64","userId":"09242844760289944401"}},"outputId":"1594e9e8-011f-4ca2-d5f5-945ee0dfe271"},"source":["# analyze clusters and choose one that is most relatable to participation\n","\n","from pprint import pprint\n","\n","index = topic_sizes['Topic'].iloc[16]\n","\n","pprint(top_n_words[index][:12])\n","print()\n","\n","pprint([sent for sent in sent_tokenize(docs_per_topic.Doc[index+1][:1200])])"],"execution_count":null,"outputs":[{"output_type":"stream","text":["[('fine', 0.6248175503695976),\n"," ('ll', 0.2695351788197904),\n"," ('class', 0.12578362045772123),\n"," ('notes', 0.11945071036643184),\n"," ('pay', 0.11153037540521053),\n"," ('attention', 0.10828350743821967),\n"," ('study', 0.1041044490026498),\n"," ('long', 0.10052034906140814),\n"," ('homework', 0.09193331463354794),\n"," ('just', 0.08552089346726328),\n"," ('read', 0.0755588712345817),\n"," ('participate', 0.05840983447565667)]\n","\n","['As long as you do the work, participate in class, and watch the films, you '\n"," 'will do fine.',\n"," 'Summers is the man, and as long as you show up and contribute to class, '\n"," \"you'll be fine.\",\n"," 'For tests, just know the review guide and which authors said what, and '\n"," \"you'll be fine.\",\n"," 'Come to class and you will be fine.',\n"," \"Classes were lecture heavy - study them and you'll do fine.\",\n"," 'As long as you study for his slides, will be fine, meaning getting an A.',\n"," \"If you go to lecture (attendance taken) and do the readings you'll be fine.\",\n"," \"Skim the readings and come up with questions before class and you'll be \"\n"," 'fine.',\n"," 'Take this class for an easy A but read your notes!',\n"," \"Take notes from textbook and pay attention in class/discussion and you'll \"\n"," \"get an easy A Go to lecture and you'll be just fine.\",\n"," \"As long as you go to lecture and discussion (since that's where you get all \"\n"," \"your info for tests), you'll be fine.\",\n"," 'Just speak up in class sometimes and study what he tells you to for tests, '\n"," \"and you'll be fine.\",\n"," \"As long as you participate in class, you'll do fine.\",\n"," \"Talk to him about your papers and you'll be fine.\",\n"," \"Take good notes during the lectures and memorize and you'll be fine.\",\n"," 'If you follow her instructions and show up to class yo']\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"d04FELM0vju0"},"source":["# save #5 cluster above (releated to workload)\n","address = \"/content/drive/My Drive/Rebuild my Professor/Clusters_xue_Mar7/\" + \"workload2\" + \".csv\"\n","doc_index = topic_sizes['Topic'].iloc[5]\n","sentences = sent_tokenize(docs_per_topic.Doc[doc_index+1])\n","with open(address, \"w\") as f:\n","    df = pd.DataFrame(sentences)\n","    df.to_csv(address)\n","\n","# save #12 cluster above (releated to workload)\n","address = \"/content/drive/My Drive/Rebuild my Professor/Clusters_xue_Mar7/\" + \"workload3\" + \".csv\"\n","doc_index = topic_sizes['Topic'].iloc[12]\n","sentences = sent_tokenize(docs_per_topic.Doc[doc_index+1])\n","with open(address, \"w\") as f:\n","    df = pd.DataFrame(sentences)\n","    df.to_csv(address)\n","\n","\n","# save #15 cluster above (releated to participation)\n","address = \"/content/drive/My Drive/Rebuild my Professor/Clusters_xue_Mar7/\" + \"participation1\" + \".csv\"\n","doc_index = topic_sizes['Topic'].iloc[15]\n","sentences = sent_tokenize(docs_per_topic.Doc[doc_index+1])\n","with open(address, \"w\") as f:\n","    df = pd.DataFrame(sentences)\n","    df.to_csv(address)\n","\n","# save #16 cluster above (releated to participation)\n","address = \"/content/drive/My Drive/Rebuild my Professor/Clusters_xue_Mar7/\" + \"participation2\" + \".csv\"\n","doc_index = topic_sizes['Topic'].iloc[16]\n","sentences = sent_tokenize(docs_per_topic.Doc[doc_index+1])\n","with open(address, \"w\") as f:\n","    df = pd.DataFrame(sentences)\n","    df.to_csv(address)\n"],"execution_count":null,"outputs":[]}]}