feature-importance-on-interview-questions.ipynb

{"cells":[{"source":"<a href=\"https://www.kaggle.com/code/testanother/feature-importance-on-interview-questions?scriptVersionId=129438284\" target=\"_blank\"><img align=\"left\" alt=\"Kaggle\" title=\"Open in Kaggle\" src=\"https://kaggle.com/static/images/open-in-kaggle.svg\"></a>","metadata":{},"cell_type":"markdown"},{"cell_type":"code","execution_count":1,"id":"07c8f1c4","metadata":{"_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","execution":{"iopub.execute_input":"2023-05-13T20:08:20.143546Z","iopub.status.busy":"2023-05-13T20:08:20.143168Z","iopub.status.idle":"2023-05-13T20:08:21.679006Z","shell.execute_reply":"2023-05-13T20:08:21.678029Z"},"papermill":{"duration":1.552167,"end_time":"2023-05-13T20:08:21.68186","exception":false,"start_time":"2023-05-13T20:08:20.129693","status":"completed"},"tags":[]},"outputs":[],"source":["import numpy as np # linear algebra\n","import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\n","from sklearn.preprocessing import OneHotEncoder\n","from sklearn.model_selection import train_test_split\n","from xgboost import XGBClassifier\n","from sklearn.compose import make_column_transformer\n","from xgboost import plot_importance\n","import matplotlib.pyplot as plt\n","from imblearn.over_sampling import SMOTE\n","from sklearn.model_selection import train_test_split\n","from sklearn.metrics import confusion_matrix, ConfusionMatrixDisplay\n","from sklearn.pipeline import Pipeline\n","from sklearn.impute import SimpleImputer\n","from sklearn.preprocessing import StandardScaler\n","from sklearn.compose import ColumnTransformer\n","from collections import Counter"]},{"cell_type":"markdown","id":"b1763599","metadata":{"papermill":{"duration":0.00989,"end_time":"2023-05-13T20:08:21.702621","exception":false,"start_time":"2023-05-13T20:08:21.692731","status":"completed"},"tags":[]},"source":["## Purpose of the Notebook\n","\n","This notebook will go over the coding challenge data from leetcode data and try to make an analysis on the questions asked by Faang companies. \n","\n","It will do classification on asked_by_faang variable and will try to create a model and predict if an interview question will be asked by the faang companies."]},{"cell_type":"code","execution_count":2,"id":"1f5e0993","metadata":{"execution":{"iopub.execute_input":"2023-05-13T20:08:21.724889Z","iopub.status.busy":"2023-05-13T20:08:21.724015Z","iopub.status.idle":"2023-05-13T20:08:21.831801Z","shell.execute_reply":"2023-05-13T20:08:21.83046Z"},"papermill":{"duration":0.121826,"end_time":"2023-05-13T20:08:21.834452","exception":false,"start_time":"2023-05-13T20:08:21.712626","status":"completed"},"tags":[]},"outputs":[{"name":"stdout","output_type":"stream","text":["<class 'pandas.core.frame.DataFrame'>\n","RangeIndex: 1825 entries, 0 to 1824\n","Data columns (total 19 columns):\n"," #   Column             Non-Null Count  Dtype  \n","---  ------             --------------  -----  \n"," 0   id                 1825 non-null   int64  \n"," 1   title              1825 non-null   object \n"," 2   description        1825 non-null   object \n"," 3   is_premium         1825 non-null   int64  \n"," 4   difficulty         1825 non-null   object \n"," 5   solution_link      987 non-null    object \n"," 6   acceptance_rate    1825 non-null   float64\n"," 7   frequency          1825 non-null   float64\n"," 8   url                1825 non-null   object \n"," 9   discuss_count      1825 non-null   int64  \n"," 10  accepted           1825 non-null   object \n"," 11  submissions        1825 non-null   object \n"," 12  companies          1749 non-null   object \n"," 13  related_topics     1571 non-null   object \n"," 14  likes              1825 non-null   int64  \n"," 15  dislikes           1825 non-null   int64  \n"," 16  rating             1825 non-null   int64  \n"," 17  asked_by_faang     1825 non-null   int64  \n"," 18  similar_questions  745 non-null    object \n","dtypes: float64(2), int64(7), object(10)\n","memory usage: 271.0+ KB\n"]}],"source":["df = pd.read_csv('/kaggle/input/leetcode-problem-dataset/leetcode_dataset - lc.csv')\n","df.info()"]},{"cell_type":"markdown","id":"edeb1822","metadata":{"papermill":{"duration":0.010303,"end_time":"2023-05-13T20:08:21.855152","exception":false,"start_time":"2023-05-13T20:08:21.844849","status":"completed"},"tags":[]},"source":["# Preprocessing\n","\n","Lets take a look at the variables...\n","\n","## Variable to predict\n","\n","This is easy and it is the 'asked_by_faang' flag. We already know that it is an integer (1 if true, 0 if false). \n","\n","Lets take a look at the distribution of the variable (how many entries etc)..."]},{"cell_type":"code","execution_count":3,"id":"cb5edd31","metadata":{"execution":{"iopub.execute_input":"2023-05-13T20:08:21.879003Z","iopub.status.busy":"2023-05-13T20:08:21.878202Z","iopub.status.idle":"2023-05-13T20:08:21.887857Z","shell.execute_reply":"2023-05-13T20:08:21.886749Z"},"papermill":{"duration":0.024299,"end_time":"2023-05-13T20:08:21.890189","exception":false,"start_time":"2023-05-13T20:08:21.86589","status":"completed"},"tags":[]},"outputs":[{"data":{"text/plain":["1    1225\n","0     600\n","Name: asked_by_faang, dtype: int64"]},"execution_count":3,"metadata":{},"output_type":"execute_result"}],"source":["df.asked_by_faang.value_counts()"]},{"cell_type":"code","execution_count":4,"id":"01887703","metadata":{"execution":{"iopub.execute_input":"2023-05-13T20:08:21.91409Z","iopub.status.busy":"2023-05-13T20:08:21.913347Z","iopub.status.idle":"2023-05-13T20:08:22.063997Z","shell.execute_reply":"2023-05-13T20:08:22.062778Z"},"papermill":{"duration":0.166791,"end_time":"2023-05-13T20:08:22.067967","exception":false,"start_time":"2023-05-13T20:08:21.901176","status":"completed"},"tags":[]},"outputs":[{"data":{"text/plain":["([<matplotlib.patches.Wedge at 0x7ae0a4dff250>,\n","  <matplotlib.patches.Wedge at 0x7ae0a4dffb90>],\n"," [Text(-0.5636085606106039, 0.9446403497662184, 'Asked by faang'),\n","  Text(0.5636085606106042, -0.9446403497662182, 'Not asked by faang')],\n"," [Text(-0.3074228512421476, 0.5152583725997554, '67.1%'),\n","  Text(0.3074228512421477, -0.5152583725997554, '32.9%')])"]},"execution_count":4,"metadata":{},"output_type":"execute_result"},{"data":{"image/png":"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\n","text/plain":["<Figure size 640x480 with 1 Axes>"]},"metadata":{},"output_type":"display_data"}],"source":["plt.pie(df.asked_by_faang.value_counts(), labels=['Asked by faang', 'Not asked by faang'], autopct='%1.1f%%')"]},{"cell_type":"markdown","id":"c47c6c8c","metadata":{"papermill":{"duration":0.022844,"end_time":"2023-05-13T20:08:22.114717","exception":false,"start_time":"2023-05-13T20:08:22.091873","status":"completed"},"tags":[]},"source":["### Intuition\n","\n","The classification is imbalanced. Meaning that we will have to oversample or undersample the training data\n","\n","## Null Values\n","\n","We also need to see if we have null values to see if we have to handle them or not."]},{"cell_type":"code","execution_count":5,"id":"e19fb359","metadata":{"execution":{"iopub.execute_input":"2023-05-13T20:08:22.141287Z","iopub.status.busy":"2023-05-13T20:08:22.140869Z","iopub.status.idle":"2023-05-13T20:08:22.152979Z","shell.execute_reply":"2023-05-13T20:08:22.152189Z"},"papermill":{"duration":0.026459,"end_time":"2023-05-13T20:08:22.155133","exception":false,"start_time":"2023-05-13T20:08:22.128674","status":"completed"},"tags":[]},"outputs":[{"data":{"text/plain":["id                      0\n","title                   0\n","description             0\n","is_premium              0\n","difficulty              0\n","solution_link         838\n","acceptance_rate         0\n","frequency               0\n","url                     0\n","discuss_count           0\n","accepted                0\n","submissions             0\n","companies              76\n","related_topics        254\n","likes                   0\n","dislikes                0\n","rating                  0\n","asked_by_faang          0\n","similar_questions    1080\n","dtype: int64"]},"execution_count":5,"metadata":{},"output_type":"execute_result"}],"source":["df.isnull().sum()"]},{"cell_type":"markdown","id":"0dad40a3","metadata":{"papermill":{"duration":0.010581,"end_time":"2023-05-13T20:08:22.176589","exception":false,"start_time":"2023-05-13T20:08:22.166008","status":"completed"},"tags":[]},"source":["Columns solution_link, companies, related_topics and similar_questions seems to have null values. \n","\n","The \"companies\" field is not interesting because we are trying to see if a question is asked by faang company so we will drop it along with id, url, solution_link and url fields. \n","\n","We also do not need the description field. It will be helpful if we want to, maybe, create a language model but there is no need to have it now.\n","\n","The field \"related questions\" however is interesting. We can create a category of questions from that one."]},{"cell_type":"markdown","id":"be1ba97e","metadata":{"papermill":{"duration":0.011124,"end_time":"2023-05-13T20:08:22.198327","exception":false,"start_time":"2023-05-13T20:08:22.187203","status":"completed"},"tags":[]},"source":["## Related Topics\n","\n","The related topics field is interesting because it is detailing which topics the questions was about. So dropping it would mean to loose lots of data. \n","\n","But we can convert the data to categorical data and treat it as such so here is what we will do: \n","\n","1. Gather all the topics so we have a list that is for example \\[topic1, topic2 ...\\]\n","2. Create columns and add them to the dataset\n","3. If the related_topics column has that topic, set the value to 1"]},{"cell_type":"code","execution_count":6,"id":"7b8c3794","metadata":{"execution":{"iopub.execute_input":"2023-05-13T20:08:22.222437Z","iopub.status.busy":"2023-05-13T20:08:22.221758Z","iopub.status.idle":"2023-05-13T20:08:22.229967Z","shell.execute_reply":"2023-05-13T20:08:22.229013Z"},"papermill":{"duration":0.023114,"end_time":"2023-05-13T20:08:22.23222","exception":false,"start_time":"2023-05-13T20:08:22.209106","status":"completed"},"tags":[]},"outputs":[],"source":["def find_unique_related_topic_values(topics):\n","    res = set()\n","    for t in topics:\n","        split = t.split(',')\n","        if len(split) != 2:\n","            continue\n","        res.add(split[0])\n","        res.add(split[1])\n","    return res\n","\n","def create_columns(data, cols):\n","    for c in cols:\n","        data[c] = 0\n","    return data\n","\n","def set_related_topic_cols(data, cols, topic_name='related_topics'):\n","    for c in cols:\n","        data[c] = data[topic_name].fillna('').apply(lambda x: 1 if c in x else 0)"]},{"cell_type":"code","execution_count":7,"id":"4f29c303","metadata":{"execution":{"iopub.execute_input":"2023-05-13T20:08:22.255697Z","iopub.status.busy":"2023-05-13T20:08:22.255054Z","iopub.status.idle":"2023-05-13T20:08:22.264327Z","shell.execute_reply":"2023-05-13T20:08:22.26327Z"},"papermill":{"duration":0.023511,"end_time":"2023-05-13T20:08:22.266764","exception":false,"start_time":"2023-05-13T20:08:22.243253","status":"completed"},"tags":[]},"outputs":[{"data":{"text/plain":["{'Array',\n"," 'Backtracking',\n"," 'Binary Indexed Tree',\n"," 'Binary Search',\n"," 'Binary Search Tree',\n"," 'Bit Manipulation',\n"," 'Brainteaser',\n"," 'Breadth-first Search',\n"," 'Depth-first Search',\n"," 'Design',\n"," 'Divide and Conquer',\n"," 'Dynamic Programming',\n"," 'Geometry',\n"," 'Graph',\n"," 'Greedy',\n"," 'Hash Table',\n"," 'Heap',\n"," 'Line Sweep',\n"," 'Linked List',\n"," 'Math',\n"," 'Meet in the Middle',\n"," 'Minimax',\n"," 'Ordered Map',\n"," 'Queue',\n"," 'Random',\n"," 'Recursion',\n"," 'Rejection Sampling',\n"," 'Rolling Hash',\n"," 'Segment Tree',\n"," 'Sliding Window',\n"," 'Sort',\n"," 'Stack',\n"," 'String',\n"," 'Suffix Array',\n"," 'Topological Sort',\n"," 'Tree',\n"," 'Trie',\n"," 'Two Pointers',\n"," 'Union Find'}"]},"execution_count":7,"metadata":{},"output_type":"execute_result"}],"source":["related_question_topics = find_unique_related_topic_values(df['related_topics'].fillna(''))\n","related_question_topics"]},{"cell_type":"code","execution_count":8,"id":"090b9603","metadata":{"execution":{"iopub.execute_input":"2023-05-13T20:08:22.290868Z","iopub.status.busy":"2023-05-13T20:08:22.290505Z","iopub.status.idle":"2023-05-13T20:08:22.295384Z","shell.execute_reply":"2023-05-13T20:08:22.294497Z"},"papermill":{"duration":0.019725,"end_time":"2023-05-13T20:08:22.29783","exception":false,"start_time":"2023-05-13T20:08:22.278105","status":"completed"},"tags":[]},"outputs":[],"source":["# It is always a good idea to copy the dataframe in case things go bad\n","df_copy = df.copy()"]},{"cell_type":"code","execution_count":9,"id":"def8d6ed","metadata":{"execution":{"iopub.execute_input":"2023-05-13T20:08:22.321176Z","iopub.status.busy":"2023-05-13T20:08:22.320871Z","iopub.status.idle":"2023-05-13T20:08:22.408078Z","shell.execute_reply":"2023-05-13T20:08:22.406864Z"},"papermill":{"duration":0.102172,"end_time":"2023-05-13T20:08:22.410907","exception":false,"start_time":"2023-05-13T20:08:22.308735","status":"completed"},"tags":[]},"outputs":[],"source":["create_columns(df_copy, related_question_topics)\n","set_related_topic_cols(df_copy, related_question_topics)"]},{"cell_type":"code","execution_count":10,"id":"978249b2","metadata":{"execution":{"iopub.execute_input":"2023-05-13T20:08:22.436017Z","iopub.status.busy":"2023-05-13T20:08:22.435665Z","iopub.status.idle":"2023-05-13T20:08:22.46218Z","shell.execute_reply":"2023-05-13T20:08:22.460554Z"},"papermill":{"duration":0.041644,"end_time":"2023-05-13T20:08:22.464685","exception":false,"start_time":"2023-05-13T20:08:22.423041","status":"completed"},"tags":[]},"outputs":[{"name":"stdout","output_type":"stream","text":["<class 'pandas.core.frame.DataFrame'>\n","RangeIndex: 1825 entries, 0 to 1824\n","Data columns (total 58 columns):\n"," #   Column                Non-Null Count  Dtype  \n","---  ------                --------------  -----  \n"," 0   id                    1825 non-null   int64  \n"," 1   title                 1825 non-null   object \n"," 2   description           1825 non-null   object \n"," 3   is_premium            1825 non-null   int64  \n"," 4   difficulty            1825 non-null   object \n"," 5   solution_link         987 non-null    object \n"," 6   acceptance_rate       1825 non-null   float64\n"," 7   frequency             1825 non-null   float64\n"," 8   url                   1825 non-null   object \n"," 9   discuss_count         1825 non-null   int64  \n"," 10  accepted              1825 non-null   object \n"," 11  submissions           1825 non-null   object \n"," 12  companies             1749 non-null   object \n"," 13  related_topics        1571 non-null   object \n"," 14  likes                 1825 non-null   int64  \n"," 15  dislikes              1825 non-null   int64  \n"," 16  rating                1825 non-null   int64  \n"," 17  asked_by_faang        1825 non-null   int64  \n"," 18  similar_questions     745 non-null    object \n"," 19  Depth-first Search    1825 non-null   int64  \n"," 20  Binary Indexed Tree   1825 non-null   int64  \n"," 21  Recursion             1825 non-null   int64  \n"," 22  Array                 1825 non-null   int64  \n"," 23  Linked List           1825 non-null   int64  \n"," 24  Divide and Conquer    1825 non-null   int64  \n"," 25  Graph                 1825 non-null   int64  \n"," 26  Geometry              1825 non-null   int64  \n"," 27  Binary Search Tree    1825 non-null   int64  \n"," 28  Sort                  1825 non-null   int64  \n"," 29  Tree                  1825 non-null   int64  \n"," 30  Union Find            1825 non-null   int64  \n"," 31  Segment Tree          1825 non-null   int64  \n"," 32  Meet in the Middle    1825 non-null   int64  \n"," 33  Trie                  1825 non-null   int64  \n"," 34  Greedy                1825 non-null   int64  \n"," 35  Minimax               1825 non-null   int64  \n"," 36  Hash Table            1825 non-null   int64  \n"," 37  Queue                 1825 non-null   int64  \n"," 38  String                1825 non-null   int64  \n"," 39  Heap                  1825 non-null   int64  \n"," 40  Bit Manipulation      1825 non-null   int64  \n"," 41  Stack                 1825 non-null   int64  \n"," 42  Breadth-first Search  1825 non-null   int64  \n"," 43  Math                  1825 non-null   int64  \n"," 44  Two Pointers          1825 non-null   int64  \n"," 45  Brainteaser           1825 non-null   int64  \n"," 46  Suffix Array          1825 non-null   int64  \n"," 47  Rolling Hash          1825 non-null   int64  \n"," 48  Sliding Window        1825 non-null   int64  \n"," 49  Design                1825 non-null   int64  \n"," 50  Random                1825 non-null   int64  \n"," 51  Dynamic Programming   1825 non-null   int64  \n"," 52  Ordered Map           1825 non-null   int64  \n"," 53  Rejection Sampling    1825 non-null   int64  \n"," 54  Backtracking          1825 non-null   int64  \n"," 55  Binary Search         1825 non-null   int64  \n"," 56  Topological Sort      1825 non-null   int64  \n"," 57  Line Sweep            1825 non-null   int64  \n","dtypes: float64(2), int64(46), object(10)\n","memory usage: 827.1+ KB\n"]}],"source":["df_copy.info()"]},{"cell_type":"markdown","id":"58c43d09","metadata":{"papermill":{"duration":0.011865,"end_time":"2023-05-13T20:08:22.48845","exception":false,"start_time":"2023-05-13T20:08:22.476585","status":"completed"},"tags":[]},"source":["So now we have some more columns which can tell us what topic was the question related to. \n","\n","## Converting the numbers represented as strings \n","\n","There are entries such as 1.2M at the accepted and submissions columns. We need to represent them as number too\n","\n","I have taken the conversion from https://stackoverflow.com/questions/39684548/convert-the-string-2-90k-to-2900-or-5-2m-to-5200000-in-pandas-dataframe"]},{"cell_type":"code","execution_count":11,"id":"33b7139c","metadata":{"execution":{"iopub.execute_input":"2023-05-13T20:08:22.514303Z","iopub.status.busy":"2023-05-13T20:08:22.513869Z","iopub.status.idle":"2023-05-13T20:08:22.520731Z","shell.execute_reply":"2023-05-13T20:08:22.519691Z"},"papermill":{"duration":0.022234,"end_time":"2023-05-13T20:08:22.522921","exception":false,"start_time":"2023-05-13T20:08:22.500687","status":"completed"},"tags":[]},"outputs":[],"source":["def convert_string_to_numeric(data, columns=['accepted', 'submissions']):\n","    for c in columns:\n","        df_copy[c] = df[c].replace(r'[KM]+$', '', regex=True).astype(float) * df[c].str.extract(r'[\\d\\.]+([KM]+)', expand=False).fillna(1).replace(['K','M'], [10**3, 10**6]).astype(int)"]},{"cell_type":"code","execution_count":12,"id":"4094d45c","metadata":{"execution":{"iopub.execute_input":"2023-05-13T20:08:22.547276Z","iopub.status.busy":"2023-05-13T20:08:22.546897Z","iopub.status.idle":"2023-05-13T20:08:22.601081Z","shell.execute_reply":"2023-05-13T20:08:22.599736Z"},"papermill":{"duration":0.069345,"end_time":"2023-05-13T20:08:22.603806","exception":false,"start_time":"2023-05-13T20:08:22.534461","status":"completed"},"tags":[]},"outputs":[{"name":"stdout","output_type":"stream","text":["<class 'pandas.core.frame.DataFrame'>\n","RangeIndex: 1825 entries, 0 to 1824\n","Data columns (total 58 columns):\n"," #   Column                Non-Null Count  Dtype  \n","---  ------                --------------  -----  \n"," 0   id                    1825 non-null   int64  \n"," 1   title                 1825 non-null   object \n"," 2   description           1825 non-null   object \n"," 3   is_premium            1825 non-null   int64  \n"," 4   difficulty            1825 non-null   object \n"," 5   solution_link         987 non-null    object \n"," 6   acceptance_rate       1825 non-null   float64\n"," 7   frequency             1825 non-null   float64\n"," 8   url                   1825 non-null   object \n"," 9   discuss_count         1825 non-null   int64  \n"," 10  accepted              1825 non-null   float64\n"," 11  submissions           1825 non-null   float64\n"," 12  companies             1749 non-null   object \n"," 13  related_topics        1571 non-null   object \n"," 14  likes                 1825 non-null   int64  \n"," 15  dislikes              1825 non-null   int64  \n"," 16  rating                1825 non-null   int64  \n"," 17  asked_by_faang        1825 non-null   int64  \n"," 18  similar_questions     745 non-null    object \n"," 19  Depth-first Search    1825 non-null   int64  \n"," 20  Binary Indexed Tree   1825 non-null   int64  \n"," 21  Recursion             1825 non-null   int64  \n"," 22  Array                 1825 non-null   int64  \n"," 23  Linked List           1825 non-null   int64  \n"," 24  Divide and Conquer    1825 non-null   int64  \n"," 25  Graph                 1825 non-null   int64  \n"," 26  Geometry              1825 non-null   int64  \n"," 27  Binary Search Tree    1825 non-null   int64  \n"," 28  Sort                  1825 non-null   int64  \n"," 29  Tree                  1825 non-null   int64  \n"," 30  Union Find            1825 non-null   int64  \n"," 31  Segment Tree          1825 non-null   int64  \n"," 32  Meet in the Middle    1825 non-null   int64  \n"," 33  Trie                  1825 non-null   int64  \n"," 34  Greedy                1825 non-null   int64  \n"," 35  Minimax               1825 non-null   int64  \n"," 36  Hash Table            1825 non-null   int64  \n"," 37  Queue                 1825 non-null   int64  \n"," 38  String                1825 non-null   int64  \n"," 39  Heap                  1825 non-null   int64  \n"," 40  Bit Manipulation      1825 non-null   int64  \n"," 41  Stack                 1825 non-null   int64  \n"," 42  Breadth-first Search  1825 non-null   int64  \n"," 43  Math                  1825 non-null   int64  \n"," 44  Two Pointers          1825 non-null   int64  \n"," 45  Brainteaser           1825 non-null   int64  \n"," 46  Suffix Array          1825 non-null   int64  \n"," 47  Rolling Hash          1825 non-null   int64  \n"," 48  Sliding Window        1825 non-null   int64  \n"," 49  Design                1825 non-null   int64  \n"," 50  Random                1825 non-null   int64  \n"," 51  Dynamic Programming   1825 non-null   int64  \n"," 52  Ordered Map           1825 non-null   int64  \n"," 53  Rejection Sampling    1825 non-null   int64  \n"," 54  Backtracking          1825 non-null   int64  \n"," 55  Binary Search         1825 non-null   int64  \n"," 56  Topological Sort      1825 non-null   int64  \n"," 57  Line Sweep            1825 non-null   int64  \n","dtypes: float64(4), int64(46), object(8)\n","memory usage: 827.1+ KB\n"]}],"source":["convert_string_to_numeric(df_copy)\n","df_copy.info()"]},{"cell_type":"markdown","id":"71b533fa","metadata":{"papermill":{"duration":0.011202,"end_time":"2023-05-13T20:08:22.626528","exception":false,"start_time":"2023-05-13T20:08:22.615326","status":"completed"},"tags":[]},"source":["## Difficulty\n","\n","Difficulty can be interpreted as categorical values. So we can use one hot encoder for sklearn"]},{"cell_type":"code","execution_count":13,"id":"25b87350","metadata":{"execution":{"iopub.execute_input":"2023-05-13T20:08:22.650444Z","iopub.status.busy":"2023-05-13T20:08:22.65008Z","iopub.status.idle":"2023-05-13T20:08:22.655887Z","shell.execute_reply":"2023-05-13T20:08:22.655071Z"},"papermill":{"duration":0.019898,"end_time":"2023-05-13T20:08:22.657849","exception":false,"start_time":"2023-05-13T20:08:22.637951","status":"completed"},"tags":[]},"outputs":[],"source":["def convert_difficulty(data, column_name='difficulty'):\n","    enc = OneHotEncoder()\n","    enc_data = pd.DataFrame(enc.fit_transform(data[[column_name]]).toarray())\n","    print(enc.get_feature_names(['difficulty']))\n","    df_encoded = data.join(enc_data)\n","    return df_encoded"]},{"cell_type":"code","execution_count":14,"id":"43b4d4c6","metadata":{"execution":{"iopub.execute_input":"2023-05-13T20:08:22.681899Z","iopub.status.busy":"2023-05-13T20:08:22.68156Z","iopub.status.idle":"2023-05-13T20:08:22.711276Z","shell.execute_reply":"2023-05-13T20:08:22.709282Z"},"papermill":{"duration":0.044864,"end_time":"2023-05-13T20:08:22.713868","exception":false,"start_time":"2023-05-13T20:08:22.669004","status":"completed"},"tags":[]},"outputs":[{"name":"stdout","output_type":"stream","text":["['difficulty_Easy' 'difficulty_Hard' 'difficulty_Medium']\n","<class 'pandas.core.frame.DataFrame'>\n","RangeIndex: 1825 entries, 0 to 1824\n","Data columns (total 61 columns):\n"," #   Column                Non-Null Count  Dtype  \n","---  ------                --------------  -----  \n"," 0   id                    1825 non-null   int64  \n"," 1   title                 1825 non-null   object \n"," 2   description           1825 non-null   object \n"," 3   is_premium            1825 non-null   int64  \n"," 4   difficulty            1825 non-null   object \n"," 5   solution_link         987 non-null    object \n"," 6   acceptance_rate       1825 non-null   float64\n"," 7   frequency             1825 non-null   float64\n"," 8   url                   1825 non-null   object \n"," 9   discuss_count         1825 non-null   int64  \n"," 10  accepted              1825 non-null   float64\n"," 11  submissions           1825 non-null   float64\n"," 12  companies             1749 non-null   object \n"," 13  related_topics        1571 non-null   object \n"," 14  likes                 1825 non-null   int64  \n"," 15  dislikes              1825 non-null   int64  \n"," 16  rating                1825 non-null   int64  \n"," 17  asked_by_faang        1825 non-null   int64  \n"," 18  similar_questions     745 non-null    object \n"," 19  Depth-first Search    1825 non-null   int64  \n"," 20  Binary Indexed Tree   1825 non-null   int64  \n"," 21  Recursion             1825 non-null   int64  \n"," 22  Array                 1825 non-null   int64  \n"," 23  Linked List           1825 non-null   int64  \n"," 24  Divide and Conquer    1825 non-null   int64  \n"," 25  Graph                 1825 non-null   int64  \n"," 26  Geometry              1825 non-null   int64  \n"," 27  Binary Search Tree    1825 non-null   int64  \n"," 28  Sort                  1825 non-null   int64  \n"," 29  Tree                  1825 non-null   int64  \n"," 30  Union Find            1825 non-null   int64  \n"," 31  Segment Tree          1825 non-null   int64  \n"," 32  Meet in the Middle    1825 non-null   int64  \n"," 33  Trie                  1825 non-null   int64  \n"," 34  Greedy                1825 non-null   int64  \n"," 35  Minimax               1825 non-null   int64  \n"," 36  Hash Table            1825 non-null   int64  \n"," 37  Queue                 1825 non-null   int64  \n"," 38  String                1825 non-null   int64  \n"," 39  Heap                  1825 non-null   int64  \n"," 40  Bit Manipulation      1825 non-null   int64  \n"," 41  Stack                 1825 non-null   int64  \n"," 42  Breadth-first Search  1825 non-null   int64  \n"," 43  Math                  1825 non-null   int64  \n"," 44  Two Pointers          1825 non-null   int64  \n"," 45  Brainteaser           1825 non-null   int64  \n"," 46  Suffix Array          1825 non-null   int64  \n"," 47  Rolling Hash          1825 non-null   int64  \n"," 48  Sliding Window        1825 non-null   int64  \n"," 49  Design                1825 non-null   int64  \n"," 50  Random                1825 non-null   int64  \n"," 51  Dynamic Programming   1825 non-null   int64  \n"," 52  Ordered Map           1825 non-null   int64  \n"," 53  Rejection Sampling    1825 non-null   int64  \n"," 54  Backtracking          1825 non-null   int64  \n"," 55  Binary Search         1825 non-null   int64  \n"," 56  Topological Sort      1825 non-null   int64  \n"," 57  Line Sweep            1825 non-null   int64  \n"," 58  0                     1825 non-null   float64\n"," 59  1                     1825 non-null   float64\n"," 60  2                     1825 non-null   float64\n","dtypes: float64(7), int64(46), object(8)\n","memory usage: 869.9+ KB\n"]},{"name":"stderr","output_type":"stream","text":["/opt/conda/lib/python3.7/site-packages/sklearn/utils/deprecation.py:87: FutureWarning: Function get_feature_names is deprecated; get_feature_names is deprecated in 1.0 and will be removed in 1.2. Please use get_feature_names_out instead.\n","  warnings.warn(msg, category=FutureWarning)\n"]}],"source":["df_encoded = convert_difficulty(df_copy)\n","df_encoded.info()"]},{"cell_type":"markdown","id":"9905ed4c","metadata":{"papermill":{"duration":0.011435,"end_time":"2023-05-13T20:08:22.73737","exception":false,"start_time":"2023-05-13T20:08:22.725935","status":"completed"},"tags":[]},"source":["So Easy is 0, Hard is 1 and medium is 2"]},{"cell_type":"markdown","id":"72c3df90","metadata":{"papermill":{"duration":0.011812,"end_time":"2023-05-13T20:08:22.760686","exception":false,"start_time":"2023-05-13T20:08:22.748874","status":"completed"},"tags":[]},"source":["## Dropping the unnecessary columns \n","\n","Now we can simply remove the columns that is not necessary"]},{"cell_type":"code","execution_count":15,"id":"3ca7b831","metadata":{"execution":{"iopub.execute_input":"2023-05-13T20:08:22.786345Z","iopub.status.busy":"2023-05-13T20:08:22.785562Z","iopub.status.idle":"2023-05-13T20:08:22.7947Z","shell.execute_reply":"2023-05-13T20:08:22.793488Z"},"papermill":{"duration":0.024596,"end_time":"2023-05-13T20:08:22.797146","exception":false,"start_time":"2023-05-13T20:08:22.77255","status":"completed"},"tags":[]},"outputs":[],"source":["drop_columns = ['id', 'title', 'description', 'difficulty', 'solution_link', 'url', 'companies', 'related_topics', 'similar_questions']\n","df_encoded.drop(columns=drop_columns, inplace=True)"]},{"cell_type":"markdown","id":"0bcaf02d","metadata":{"papermill":{"duration":0.011679,"end_time":"2023-05-13T20:08:22.820924","exception":false,"start_time":"2023-05-13T20:08:22.809245","status":"completed"},"tags":[]},"source":["# Modeling \n","\n","This process is straigtforward. I will split the data into train and test and then train a classifier"]},{"cell_type":"code","execution_count":16,"id":"8186b517","metadata":{"execution":{"iopub.execute_input":"2023-05-13T20:08:22.84582Z","iopub.status.busy":"2023-05-13T20:08:22.845454Z","iopub.status.idle":"2023-05-13T20:08:22.854128Z","shell.execute_reply":"2023-05-13T20:08:22.852901Z"},"papermill":{"duration":0.024039,"end_time":"2023-05-13T20:08:22.85648","exception":false,"start_time":"2023-05-13T20:08:22.832441","status":"completed"},"tags":[]},"outputs":[],"source":["X, y = df_encoded.drop(columns=['asked_by_faang']), df_encoded['asked_by_faang']\n","\n","X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.33, random_state=42)"]},{"cell_type":"code","execution_count":17,"id":"7bdaef4e","metadata":{"execution":{"iopub.execute_input":"2023-05-13T20:08:22.884564Z","iopub.status.busy":"2023-05-13T20:08:22.883917Z","iopub.status.idle":"2023-05-13T20:08:22.915263Z","shell.execute_reply":"2023-05-13T20:08:22.914463Z"},"papermill":{"duration":0.048404,"end_time":"2023-05-13T20:08:22.91782","exception":false,"start_time":"2023-05-13T20:08:22.869416","status":"completed"},"tags":[]},"outputs":[{"name":"stderr","output_type":"stream","text":["/opt/conda/lib/python3.7/site-packages/sklearn/utils/validation.py:1692: FutureWarning: Feature names only support names that are all strings. Got feature names with dtypes: ['int', 'str']. An error will be raised in 1.2.\n","  FutureWarning,\n"]}],"source":["# scale the train set up to have a balance classification\n","sm = SMOTE(random_state=42)\n","X_res, y_res = sm.fit_resample(X_train, y_train)"]},{"cell_type":"code","execution_count":18,"id":"ad726dfd","metadata":{"execution":{"iopub.execute_input":"2023-05-13T20:08:22.951746Z","iopub.status.busy":"2023-05-13T20:08:22.950051Z","iopub.status.idle":"2023-05-13T20:08:22.961049Z","shell.execute_reply":"2023-05-13T20:08:22.959605Z"},"papermill":{"duration":0.033781,"end_time":"2023-05-13T20:08:22.963622","exception":false,"start_time":"2023-05-13T20:08:22.929841","status":"completed"},"tags":[]},"outputs":[{"data":{"text/plain":["Counter({1: 834, 0: 834})"]},"execution_count":18,"metadata":{},"output_type":"execute_result"}],"source":["# See if it worked\n","Counter(y_res)"]},{"cell_type":"code","execution_count":19,"id":"e6fdb855","metadata":{"execution":{"iopub.execute_input":"2023-05-13T20:08:23.020736Z","iopub.status.busy":"2023-05-13T20:08:23.019455Z","iopub.status.idle":"2023-05-13T20:08:23.02765Z","shell.execute_reply":"2023-05-13T20:08:23.026784Z"},"papermill":{"duration":0.039855,"end_time":"2023-05-13T20:08:23.030153","exception":false,"start_time":"2023-05-13T20:08:22.990298","status":"completed"},"tags":[]},"outputs":[],"source":["numeric_features = X_res.columns\n","\n","numeric_transformer = Pipeline(steps=[\n","    ('imputer', SimpleImputer(strategy='median')),\n","    ('scaler', StandardScaler()),\n","])\n","\n","preprocessor = ColumnTransformer(\n","    transformers=[\n","        ('num', numeric_transformer, numeric_features)\n","    ]\n",")\n","\n","model = XGBClassifier(objective='binary:logistic', random_state=42)\n","\n","clf = Pipeline([\n","     ('preprocessor', preprocessor),\n","     ('clf', model)\n","])"]},{"cell_type":"code","execution_count":20,"id":"f5533f7b","metadata":{"execution":{"iopub.execute_input":"2023-05-13T20:08:23.058774Z","iopub.status.busy":"2023-05-13T20:08:23.057982Z","iopub.status.idle":"2023-05-13T20:08:23.739154Z","shell.execute_reply":"2023-05-13T20:08:23.737898Z"},"papermill":{"duration":0.6987,"end_time":"2023-05-13T20:08:23.741989","exception":false,"start_time":"2023-05-13T20:08:23.043289","status":"completed"},"tags":[]},"outputs":[{"name":"stderr","output_type":"stream","text":["/opt/conda/lib/python3.7/site-packages/sklearn/utils/validation.py:1692: FutureWarning: Feature names only support names that are all strings. Got feature names with dtypes: ['int', 'str']. An error will be raised in 1.2.\n","  FutureWarning,\n","/opt/conda/lib/python3.7/site-packages/sklearn/utils/validation.py:1692: FutureWarning: Feature names only support names that are all strings. Got feature names with dtypes: ['int', 'str']. An error will be raised in 1.2.\n","  FutureWarning,\n"]},{"data":{"text/plain":["Pipeline(steps=[('preprocessor',\n","                 ColumnTransformer(transformers=[('num',\n","                                                  Pipeline(steps=[('imputer',\n","                                                                   SimpleImputer(strategy='median')),\n","                                                                  ('scaler',\n","                                                                   StandardScaler())]),\n","                                                  Index([          'is_premium',      'acceptance_rate',            'frequency',\n","              'discuss_count',             'accepted',          'submissions',\n","                      'likes',             'dislikes',               'rating',\n","         'Depth-first Search',  'Binary Indexed Tree',            'Recursion',\n","                      'Array',          'Linked Li...\n","                               gamma=0, gpu_id=-1, grow_policy='depthwise',\n","                               importance_type=None, interaction_constraints='',\n","                               learning_rate=0.300000012, max_bin=256,\n","                               max_cat_to_onehot=4, max_delta_step=0,\n","                               max_depth=6, max_leaves=0, min_child_weight=1,\n","                               missing=nan, monotone_constraints='()',\n","                               n_estimators=100, n_jobs=0, num_parallel_tree=1,\n","                               predictor='auto', random_state=42, reg_alpha=0,\n","                               reg_lambda=1, ...))])"]},"execution_count":20,"metadata":{},"output_type":"execute_result"}],"source":["clf.fit(X_res, y_res)"]},{"cell_type":"code","execution_count":21,"id":"58b6bdd2","metadata":{"execution":{"iopub.execute_input":"2023-05-13T20:08:23.770384Z","iopub.status.busy":"2023-05-13T20:08:23.76956Z","iopub.status.idle":"2023-05-13T20:08:23.791624Z","shell.execute_reply":"2023-05-13T20:08:23.790613Z"},"papermill":{"duration":0.039803,"end_time":"2023-05-13T20:08:23.79475","exception":false,"start_time":"2023-05-13T20:08:23.754947","status":"completed"},"tags":[]},"outputs":[{"name":"stderr","output_type":"stream","text":["/opt/conda/lib/python3.7/site-packages/sklearn/utils/validation.py:1692: FutureWarning: Feature names only support names that are all strings. Got feature names with dtypes: ['int', 'str']. An error will be raised in 1.2.\n","  FutureWarning,\n"]},{"data":{"text/plain":["0.6666666666666666"]},"execution_count":21,"metadata":{},"output_type":"execute_result"}],"source":["clf.score(X_test, y_test)"]},{"cell_type":"markdown","id":"b485cc03","metadata":{"papermill":{"duration":0.012604,"end_time":"2023-05-13T20:08:23.821006","exception":false,"start_time":"2023-05-13T20:08:23.808402","status":"completed"},"tags":[]},"source":["The model has a 66% accuracy. This is not bad but could be way better. \n","\n","We can do randomized search to find better parameters for the classfier too but that is future work for now"]},{"cell_type":"code","execution_count":22,"id":"012c27c8","metadata":{"execution":{"iopub.execute_input":"2023-05-13T20:08:23.849098Z","iopub.status.busy":"2023-05-13T20:08:23.848022Z","iopub.status.idle":"2023-05-13T20:08:23.865414Z","shell.execute_reply":"2023-05-13T20:08:23.864361Z"},"papermill":{"duration":0.03572,"end_time":"2023-05-13T20:08:23.869894","exception":false,"start_time":"2023-05-13T20:08:23.834174","status":"completed"},"tags":[]},"outputs":[{"name":"stderr","output_type":"stream","text":["/opt/conda/lib/python3.7/site-packages/sklearn/utils/validation.py:1692: FutureWarning: Feature names only support names that are all strings. Got feature names with dtypes: ['int', 'str']. An error will be raised in 1.2.\n","  FutureWarning,\n"]}],"source":["predictions = clf.predict(X_test)\n","cm = confusion_matrix(y_test, predictions, labels=clf.classes_)"]},{"cell_type":"code","execution_count":23,"id":"7e1319c0","metadata":{"execution":{"iopub.execute_input":"2023-05-13T20:08:23.897464Z","iopub.status.busy":"2023-05-13T20:08:23.897182Z","iopub.status.idle":"2023-05-13T20:08:24.171871Z","shell.execute_reply":"2023-05-13T20:08:24.170674Z"},"papermill":{"duration":0.29055,"end_time":"2023-05-13T20:08:24.174383","exception":false,"start_time":"2023-05-13T20:08:23.883833","status":"completed"},"tags":[]},"outputs":[{"data":{"text/plain":["<sklearn.metrics._plot.confusion_matrix.ConfusionMatrixDisplay at 0x7ae09b51bdd0>"]},"execution_count":23,"metadata":{},"output_type":"execute_result"},{"data":{"image/png":"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\n","text/plain":["<Figure size 640x480 with 2 Axes>"]},"metadata":{},"output_type":"display_data"}],"source":["disp = ConfusionMatrixDisplay(confusion_matrix=cm, display_labels=clf.classes_)\n","disp.plot()"]},{"cell_type":"markdown","id":"50ad5ca9","metadata":{"papermill":{"duration":0.012889,"end_time":"2023-05-13T20:08:24.200505","exception":false,"start_time":"2023-05-13T20:08:24.187616","status":"completed"},"tags":[]},"source":["The model is not that great at identifying the true negatives (112 entries were not asked by faang companies but was predicted as asked by faang companies\n","\n","# Feature Importance\n","\n","We now can use feature importance on the trained model. I will use the SHAP library https://shap.readthedocs.io/en/latest/index.html"]},{"cell_type":"code","execution_count":24,"id":"227b743a","metadata":{"execution":{"iopub.execute_input":"2023-05-13T20:08:24.228349Z","iopub.status.busy":"2023-05-13T20:08:24.228013Z","iopub.status.idle":"2023-05-13T20:08:28.90984Z","shell.execute_reply":"2023-05-13T20:08:28.908635Z"},"papermill":{"duration":4.698721,"end_time":"2023-05-13T20:08:28.912478","exception":false,"start_time":"2023-05-13T20:08:24.213757","status":"completed"},"tags":[]},"outputs":[{"name":"stderr","output_type":"stream","text":["ntree_limit is deprecated, use `iteration_range` or model slicing instead.\n"]},{"data":{"image/png":"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\n","text/plain":["<Figure size 800x950 with 1 Axes>"]},"metadata":{},"output_type":"display_data"}],"source":["import shap\n","explainer = shap.TreeExplainer(model)\n","shap_values = explainer.shap_values(X_res)\n","shap.summary_plot(shap_values, X_res, plot_type=\"bar\")"]},{"cell_type":"markdown","id":"be025f18","metadata":{"papermill":{"duration":0.014524,"end_time":"2023-05-13T20:08:28.941522","exception":false,"start_time":"2023-05-13T20:08:28.926998","status":"completed"},"tags":[]},"source":["# Final Thoughts \n","\n","So there it is. We have trained a model and made feature analysis on it. It seems like submissions, likes and frequency effects the probability of a question being asked by a faang company. This is not surprising since I assume people just try to solve questions that are asked by large tech companies. \n","\n","However if you want to practice, concentrate on questions that are easy (0 is the feature name) and hard (1 is the feature name) since they also have an effect on the result. \n","\n","# Improvement Ideas\n","\n","1. Code should be more modularized. I have omitted doing this but it can be done to make things easier for the future use \n","2. RandomizedSearch on the model. You can try this and see what the result becomes"]},{"cell_type":"code","execution_count":null,"id":"965e9c5a","metadata":{"papermill":{"duration":0.013598,"end_time":"2023-05-13T20:08:28.969683","exception":false,"start_time":"2023-05-13T20:08:28.956085","status":"completed"},"tags":[]},"outputs":[],"source":[]}],"metadata":{"kernelspec":{"display_name":"Python 3","language":"python","name":"python3"},"language_info":{"codemirror_mode":{"name":"ipython","version":3},"file_extension":".py","mimetype":"text/x-python","name":"python","nbconvert_exporter":"python","pygments_lexer":"ipython3","version":"3.7.12"},"papermill":{"default_parameters":{},"duration":21.097819,"end_time":"2023-05-13T20:08:30.006831","environment_variables":{},"exception":null,"input_path":"__notebook__.ipynb","output_path":"__notebook__.ipynb","parameters":{},"start_time":"2023-05-13T20:08:08.909012","version":"2.4.0"}},"nbformat":4,"nbformat_minor":5}