nlp-using-bert.ipynb

{"cells":[{"metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true},"cell_type":"code","source":"import numpy as np # linear algebra\nimport pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)","execution_count":1,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"import seaborn as sns\nimport scipy\nimport matplotlib.pyplot as plt\nimport scikitplot as skplt\nfrom wordcloud import WordCloud","execution_count":2,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"from bs4 import BeautifulSoup\nimport string\nimport nltk\nfrom nltk.corpus import stopwords\nfrom nltk.tokenize import sent_tokenize\nfrom nltk.tokenize import RegexpTokenizer\nfrom nltk.stem import WordNetLemmatizer\nfrom nltk.stem.porter import PorterStemmer","execution_count":3,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"from sklearn.feature_extraction.text import CountVectorizer\nfrom sklearn.feature_extraction.text import TfidfVectorizer\nfrom sklearn.feature_extraction.text import TfidfTransformer","execution_count":4,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"!wget --quiet https://raw.githubusercontent.com/tensorflow/models/master/official/nlp/bert/tokenization.py","execution_count":5,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"import tensorflow as tf\nfrom tensorflow.keras.layers import Dense, Input\nfrom tensorflow.keras.optimizers import Adam\nfrom tensorflow.keras.models import Model\nfrom tensorflow.keras.callbacks import ModelCheckpoint\nimport tensorflow_hub as hub\nimport tokenization","execution_count":6,"outputs":[]},{"metadata":{"_uuid":"d629ff2d2480ee46fbb7e2d37f6b5fab8052498a","_cell_guid":"79c7e3d0-c299-4dcb-8224-4455121ee9b0","trusted":true},"cell_type":"code","source":"import os\nimport re\nfor dirname, _, filenames in os.walk('/kaggle/input'):\n    for filename in filenames:\n        print(os.path.join(dirname, filename))","execution_count":7,"outputs":[{"output_type":"stream","text":"/kaggle/input/clean-dataset-steam/test.csv\n/kaggle/input/clean-dataset-steam/train.csv\n/kaggle/input/glove42b300dtxt/glove.42B.300d.txt\n/kaggle/input/steam-recommendation-nlp-dataset/game_overview.csv\n/kaggle/input/steam-recommendation-nlp-dataset/test.csv\n/kaggle/input/steam-recommendation-nlp-dataset/train.csv\n","name":"stdout"}]},{"metadata":{"trusted":true},"cell_type":"code","source":"df = pd.read_csv('/kaggle/input/clean-dataset-steam/train.csv')\ngame_df = pd.read_csv('/kaggle/input/steam-recommendation-nlp-dataset/game_overview.csv')\ntesting_df=  pd.read_csv('/kaggle/input/clean-dataset-steam/test.csv')","execution_count":8,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def rep(text):\n    text = re.sub('♥♥♥♥',\"\",text)\n    return text\n\ndf['user_review']=df.user_review.apply(rep)\ntesting_df['user_review']=testing_df.user_review.apply(rep)","execution_count":9,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# testing_df['user_review'].sample(5)","execution_count":10,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def low(text):\n    return text.lower()\n\ndf['user_review']=df.user_review.apply(low)\ntesting_df['user_review']=testing_df.user_review.apply(low)\n\n# testing_df.head(5)","execution_count":11,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def asc(text):\n    text = re.sub('[^a-zA-Z.,]',\" \",text)\n    return text\n\ndf['user_review']=df.user_review.apply(asc)\ntesting_df['user_review']=testing_df.user_review.apply(asc)\n\n\n# testing_df.head(5)","execution_count":12,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def space(text):\n#     text = re.sub(' +', ' ', text)\n    text = \" \".join(text.split())\n    return text\n\ndf['user_review']=df.user_review.apply(space)\ntesting_df['user_review']=testing_df.user_review.apply(space)","execution_count":13,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def clean(text):\n    text = re.sub(' ll', 'will', text)\n    text = re.sub('lvl', 'level', text)\n    text = re.sub('dev', 'developer', text)\n#     text = re.sub('ll', 'will', text)\n\n    return text\n\ndf['user_review']=df.user_review.apply(clean)\ntesting_df['user_review']=testing_df.user_review.apply(clean)","execution_count":14,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"testing_df['user_review'][2445]","execution_count":15,"outputs":[{"output_type":"execute_result","execution_count":15,"data":{"text/plain":"'till level you are like wtf how can this game get so many negative reviews then surprise after level all makes sensethe game is boring and lazy, the developers dont pay attetion to the community.all this game has to offer is classes nothing more, i m not saying that the classes are bad but thats only it, a class simulator. dungeons oh boy haha, that if you like pve you should run now.dungeons can be done only times per day if you have your token active. wtf market of the game is bs, you can place max items with active token. wtf haters will say anti bot bro shut up you that is what i call bad developer.all you do in this game is grind or quests, nothing more. my friends for a game released in thats so bad that makes me cry, i know some people like that type of game but itwill fail coz its and not , if imc wants to get more people to play they need to adapt.tl dr run dont download or support this game till imc learn to do what the community wants. but ignore all i said if you want a game to grind and grind and grind and grind and hmm grind some more for new classes and maybe a good pvp. coz pve sucks.'"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# let's summarize"},{"metadata":{"trusted":true},"cell_type":"code","source":"# import networkx as nx\n# from sklearn.metrics.pairwise import cosine_similarity\n# gloveFile = \"/kaggle/input/glove42b300dtxt/glove.42B.300d.txt\"\n# dim=300","execution_count":16,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# def loadGloveModel(gloveFile):\n#     word_embeddings = {}\n#     f = open(gloveFile, encoding='utf-8')\n#     for line in f:\n#         values = line.split()\n#         word = values[0]\n#         coefs = np.asarray(values[1:], dtype='float32')\n#         word_embeddings[word] = coefs\n#     f.close()\n#     return word_embeddings\n\n# word_embeddings = loadGloveModel(gloveFile)\n# print(\"Vocab Size = \",len(word_embeddings))\n","execution_count":17,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# def summ(text):\n#     sentences = sent_tokenize(text)\n    \n#     sentence_vectors = []\n#     for i in sentences:\n#         if len(i)!=0:\n#             v = sum([word_embeddings.get(w, np.zeros((dim,))) for w in i.split()])/(len(i.split())+0.001)\n#         else:\n#             v = np.zeros((dim,))\n#         sentence_vectors.append(v)\n        \n#     sim_mat = np.zeros([len(sentences),len(sentences)])\n    \n#     for i in range(len(sentences)):\n#         for j in range(len(sentences)):\n#             if i!=j:\n#                 sim_mat[i][j] = cosine_similarity(sentence_vectors[i].reshape(1,dim),sentence_vectors[j].reshape(1,dim))[0,0]\n\n#     sim_mat = np.round(sim_mat,3)\n#     nx_graph = nx.from_numpy_array(sim_mat)\n#     scores = nx.pagerank(nx_graph)\n#     ranked_sentences = sorted(((scores[i],i)for i,s in enumerate(sentences)),reverse=True)\n#     arranged_sentences = sorted(ranked_sentences[0:int(len(sentences)*0.5)] , key=lambda x:x[1])\n# #     print(\"\\n\".join([sentences[x[1]] for x in arranged_sentences]))\n#     text = \" \".join([sentences[x[1]] for x in arranged_sentences])\n#     return text","execution_count":18,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# testing_df['user_review'][2445]","execution_count":19,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# pip install bert-extractive-summarizer","execution_count":20,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# from summarizer import Summarizer","execution_count":21,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# def summary(text):\n#     model = Summarizer()\n#     result = model(body)\n#     text = ''.join(result)\n#     return text\n\n# df['user_review']=df.user_review.apply(summary)\n# # testing_df['user_review']=testing_df.user_review.apply(clean)","execution_count":22,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# summary not working"},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"testing_df.drop([\"review_id\",\"title\",\"year\"],axis=1,inplace=True)","execution_count":23,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"testing_df.head(5)","execution_count":24,"outputs":[{"output_type":"execute_result","execution_count":24,"data":{"text/plain":"                                         user_review\n0  nice graphics, new maps, weapons and models. b...\n1  i would not recommend getting into this at its...\n2  edit i have tried playing cs go recently and i...\n3  the game is great. but the community is the wo...\n4  i thank trulyrazor for buying this for me a lo...","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>user_review</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>nice graphics, new maps, weapons and models. b...</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>i would not recommend getting into this at its...</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>edit i have tried playing cs go recently and i...</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>the game is great. but the community is the wo...</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>i thank trulyrazor for buying this for me a lo...</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"df.drop(['review_id'],axis=1,inplace=True)","execution_count":25,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"result = pd.merge(df, game_df,on='title', how='left')","execution_count":26,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"result.drop(['year'],axis=1,inplace=True)\nresult.tail(5)","execution_count":27,"outputs":[{"output_type":"execute_result","execution_count":27,"data":{"text/plain":"              title                                        user_review  \\\n17489  EverQuest II  arguably the single greatest mmorp that exists...   \n17490  EverQuest II  an older game, to be sure, but has its own cha...   \n17491  EverQuest II  when i frist started playing everquest it was ...   \n17492  EverQuest II  cool game. the only thing that really pisses m...   \n17493  EverQuest II  this game since i was a little kid, always hav...   \n\n       user_suggestion               developer               publisher  \\\n17489                1  Daybreak Game Company   Daybreak Game Company    \n17490                1  Daybreak Game Company   Daybreak Game Company    \n17491                1  Daybreak Game Company   Daybreak Game Company    \n17492                1  Daybreak Game Company   Daybreak Game Company    \n17493                1  Daybreak Game Company   Daybreak Game Company    \n\n                                                    tags  \\\n17489  ['Free to Play', 'Massively Multiplayer', 'RPG...   \n17490  ['Free to Play', 'Massively Multiplayer', 'RPG...   \n17491  ['Free to Play', 'Massively Multiplayer', 'RPG...   \n17492  ['Free to Play', 'Massively Multiplayer', 'RPG...   \n17493  ['Free to Play', 'Massively Multiplayer', 'RPG...   \n\n                                                overview  \n17489  EverQuest® II is the epitome of massively mult...  \n17490  EverQuest® II is the epitome of massively mult...  \n17491  EverQuest® II is the epitome of massively mult...  \n17492  EverQuest® II is the epitome of massively mult...  \n17493  EverQuest® II is the epitome of massively mult...  ","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>title</th>\n      <th>user_review</th>\n      <th>user_suggestion</th>\n      <th>developer</th>\n      <th>publisher</th>\n      <th>tags</th>\n      <th>overview</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>17489</th>\n      <td>EverQuest II</td>\n      <td>arguably the single greatest mmorp that exists...</td>\n      <td>1</td>\n      <td>Daybreak Game Company</td>\n      <td>Daybreak Game Company</td>\n      <td>['Free to Play', 'Massively Multiplayer', 'RPG...</td>\n      <td>EverQuest® II is the epitome of massively mult...</td>\n    </tr>\n    <tr>\n      <th>17490</th>\n      <td>EverQuest II</td>\n      <td>an older game, to be sure, but has its own cha...</td>\n      <td>1</td>\n      <td>Daybreak Game Company</td>\n      <td>Daybreak Game Company</td>\n      <td>['Free to Play', 'Massively Multiplayer', 'RPG...</td>\n      <td>EverQuest® II is the epitome of massively mult...</td>\n    </tr>\n    <tr>\n      <th>17491</th>\n      <td>EverQuest II</td>\n      <td>when i frist started playing everquest it was ...</td>\n      <td>1</td>\n      <td>Daybreak Game Company</td>\n      <td>Daybreak Game Company</td>\n      <td>['Free to Play', 'Massively Multiplayer', 'RPG...</td>\n      <td>EverQuest® II is the epitome of massively mult...</td>\n    </tr>\n    <tr>\n      <th>17492</th>\n      <td>EverQuest II</td>\n      <td>cool game. the only thing that really pisses m...</td>\n      <td>1</td>\n      <td>Daybreak Game Company</td>\n      <td>Daybreak Game Company</td>\n      <td>['Free to Play', 'Massively Multiplayer', 'RPG...</td>\n      <td>EverQuest® II is the epitome of massively mult...</td>\n    </tr>\n    <tr>\n      <th>17493</th>\n      <td>EverQuest II</td>\n      <td>this game since i was a little kid, always hav...</td>\n      <td>1</td>\n      <td>Daybreak Game Company</td>\n      <td>Daybreak Game Company</td>\n      <td>['Free to Play', 'Massively Multiplayer', 'RPG...</td>\n      <td>EverQuest® II is the epitome of massively mult...</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"result.isnull().any()","execution_count":28,"outputs":[{"output_type":"execute_result","execution_count":28,"data":{"text/plain":"title              False\nuser_review        False\nuser_suggestion    False\ndeveloper          False\npublisher          False\ntags               False\noverview           False\ndtype: bool"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"result.drop([\"overview\",\"developer\",\"publisher\"],axis=1,inplace=True)","execution_count":29,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"result.head(2)","execution_count":30,"outputs":[{"output_type":"execute_result","execution_count":30,"data":{"text/plain":"                         title  \\\n0  Spooky's Jump Scare Mansion   \n1  Spooky's Jump Scare Mansion   \n\n                                         user_review  user_suggestion  \\\n0  i m scared and hearing creepy voices. so iwill...                1   \n1  best game, more better than sam pepper s youtu...                1   \n\n                                                tags  \n0  ['Horror', 'Free to Play', 'Cute', 'First-Pers...  \n1  ['Horror', 'Free to Play', 'Cute', 'First-Pers...  ","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>title</th>\n      <th>user_review</th>\n      <th>user_suggestion</th>\n      <th>tags</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>Spooky's Jump Scare Mansion</td>\n      <td>i m scared and hearing creepy voices. so iwill...</td>\n      <td>1</td>\n      <td>['Horror', 'Free to Play', 'Cute', 'First-Pers...</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>Spooky's Jump Scare Mansion</td>\n      <td>best game, more better than sam pepper s youtu...</td>\n      <td>1</td>\n      <td>['Horror', 'Free to Play', 'Cute', 'First-Pers...</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"testing_df.head(2)","execution_count":31,"outputs":[{"output_type":"execute_result","execution_count":31,"data":{"text/plain":"                                         user_review\n0  nice graphics, new maps, weapons and models. b...\n1  i would not recommend getting into this at its...","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>user_review</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>nice graphics, new maps, weapons and models. b...</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>i would not recommend getting into this at its...</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# Text Processing"},{"metadata":{"trusted":true},"cell_type":"code","source":"def punctuation_removal(messy_str):\n    clean_list = [char for char in messy_str if char not in string.punctuation]\n    clean_str = ''.join(clean_list)\n    return clean_str","execution_count":32,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"result['user_review'] = result['user_review'].apply(punctuation_removal)\ntesting_df['user_review'] = testing_df['user_review'].apply(punctuation_removal)\n","execution_count":33,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"testing_df.user_review.str.len().mean()","execution_count":34,"outputs":[{"output_type":"execute_result","execution_count":34,"data":{"text/plain":"748.372156619018"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"result.user_review.str.len().mean()","execution_count":35,"outputs":[{"output_type":"execute_result","execution_count":35,"data":{"text/plain":"726.3807591174117"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"\n# from nltk.tokenize import sent_tokenize, word_tokenize, RegexpTokenizer\n# stop = stopwords.words('english')","execution_count":36,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# stop.remove('not')\n# stop.remove(\"don't\")","execution_count":37,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# stop_words = []\n\n# for item in stop: \n#     new_item = punctuation_removal(item)\n#     stop_words.append(new_item) \n# print(stop_words)","execution_count":38,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# def stopwords_removal(messy_str):\n#     messy_str = word_tokenize(messy_str)\n#     return [word.lower() for word in messy_str \n#             if word.lower() not in stop_words ]","execution_count":39,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# result['user_review'] = result['user_review'].apply(stopwords_removal)\n# testing_df['user_review'] = testing_df['user_review'].apply(stopwords_removal)\n","execution_count":40,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# result.head(2)\n\n","execution_count":41,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# import re\n# def drop_numbers(list_text):\n#     list_text_new = []\n#     for i in list_text:\n#         if not re.search('\\d', i):\n#             list_text_new.append(i)\n#     return ' '.join(list_text_new)\n# result['user_review'] = result['user_review'].apply(drop_numbers)\n# testing_df['user_review'] = testing_df['user_review'].apply(drop_numbers)\n","execution_count":42,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# result.head(2)","execution_count":43,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# from textblob import TextBlob\n# example = \"This game is scary and i would recommend it playing\"\n# blob = TextBlob(example)\n# blob.sentiment","execution_count":44,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"#Note that the subjectivity is still high as it is still an opinion and not a fact"},{"metadata":{"trusted":true},"cell_type":"code","source":"# result_reviews = result['user_review'] \n# testing_reviews = testing_df['user_review'] \n\n# from tqdm import tqdm\n# result_sentiments = []\n# for review in tqdm(result_reviews):\n#     blob = TextBlob(review)\n#     result_sentiments += [blob.sentiment.polarity]\n    \n    \n# from tqdm import tqdm\n# testing_sentiments = []\n# for review in tqdm(testing_reviews):\n#     blob = TextBlob(review)\n#     testing_sentiments += [blob.sentiment.polarity]\n    \n    \n# result['sentiment'] = result_sentiments\n# testing_df['sentiment'] =testing_sentiments\n\n\n\n\n# loaded_submission = pd.read_csv('/kaggle/input/submission/submission1.csv')\n# loaded_submission.tail()\n# loaded_submission['sentiment'] = testing_df['sentiment']\n\n\n\n# result.user_review[13978] = \"good game\"\n\n# result.user_review[13978]","execution_count":45,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# result.sample(20)","execution_count":46,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# Correlation with ratings"},{"metadata":{},"cell_type":"markdown","source":"Let's see if the newly calculated sentiments have any correlation with the ratings. Our hypothesis would be that reviews with postive reviews would have higher polarity values and vice-versa."},{"metadata":{"trusted":true},"cell_type":"code","source":"# np.corrcoef(result[\"user_suggestion\"], result[\"sentiment\"])","execution_count":47,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"We observe that there is indeed a positive but not very strong correlation between the sentiment of the reviews and the ratings given (0.378). Although the correlation is weak, it is still a useful feature for us to make predictions from as it isn't close to 0."},{"metadata":{},"cell_type":"markdown","source":"# Visualizing the correlation"},{"metadata":{"trusted":true},"cell_type":"code","source":"# sns.boxplot(x=np.array(result[\"user_suggestion\"]),y=np.array(result[\"sentiment\"]))\n# plt.xlabel(\"Recommended\")\n# plt.ylabel(\"Sentiment\")\n# plt.title(\"Sentiment vs Recommended\")\n# plt.show()","execution_count":48,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"Improving the sentiment analysis\nThis was a very simple intro to sentiment analysis. We note that the sentiments calculated are only weakly correlated with the ratings, suggesting that there is a lot of room for improvement. A few basic steps that can be taken to improve our sentiment calculation are:\n\nPre-processing: Clean up the text. Do stemming/lemmatization. Remove stop words\nRemove neutral sentences - Do sentiment analysis of individual sentences in a review and remove neutral ones. Then re-run the sentiment analysis on the polar sentences. This will give a more accurate sentiment score"},{"metadata":{"trusted":true},"cell_type":"code","source":"# result.drop(['title','tags'],axis=1,inplace=True)\n# result.head(2)","execution_count":49,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# result.sentiment.min()","execution_count":50,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# result.sentiment.max()","execution_count":51,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"X = result.drop('user_suggestion',axis=1)\ny = result['user_suggestion']","execution_count":52,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"X.user_review.str.len().mean()","execution_count":53,"outputs":[{"output_type":"execute_result","execution_count":53,"data":{"text/plain":"726.3807591174117"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"from sklearn.model_selection import train_test_split\nX_train, X_test, y_train, y_test = train_test_split(X, y,random_state = 42, test_size=0.25)","execution_count":54,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"print('\\n train X: {} \\n train y: {} \\n Val X: {} \\n val y: {}'.format((X_train.shape),(y_train.shape),(X_test.shape),(y_test.shape)))","execution_count":55,"outputs":[{"output_type":"stream","text":"\n train X: (13120, 3) \n train y: (13120,) \n Val X: (4374, 3) \n val y: (4374,)\n","name":"stdout"}]},{"metadata":{"trusted":true},"cell_type":"code","source":"X_train.user_review.str.len().mean()","execution_count":56,"outputs":[{"output_type":"execute_result","execution_count":56,"data":{"text/plain":"726.2573170731707"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"X_test.user_review.str.len().mean()","execution_count":57,"outputs":[{"output_type":"execute_result","execution_count":57,"data":{"text/plain":"726.7510288065844"},"metadata":{}}]},{"metadata":{},"cell_type":"markdown","source":"# BERT"},{"metadata":{"trusted":true},"cell_type":"code","source":"module_url = \"https://tfhub.dev/tensorflow/bert_en_uncased_L-24_H-1024_A-16/1\"\nbert_layer = hub.KerasLayer(module_url, trainable=True)","execution_count":58,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"vocab_file = bert_layer.resolved_object.vocab_file.asset_path.numpy()\ndo_lower_case = bert_layer.resolved_object.do_lower_case.numpy()\ntokenizer = tokenization.FullTokenizer(vocab_file, do_lower_case)","execution_count":59,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"X_train_bert = X_train.user_review\nX_test_bert = X_test.user_review\n\n","execution_count":60,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# from tqdm import tqdm\n# train_x_sentiment = []\n# for review in tqdm(X_train_bert):\n#     blob = TextBlob(review)\n#     train_x_sentiment += [blob.sentiment.polarity]","execution_count":61,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# from tqdm import tqdm\n# test_x_sentiment = []\n# for review in tqdm(X_test_bert):\n#     blob = TextBlob(review)\n#     test_x_sentiment += [blob.sentiment.polarity]","execution_count":62,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# X_train_bert","execution_count":63,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# X_train_bert.head(2)","execution_count":64,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def bert_encode(input_text, tokenizer, max_len = 512):\n    token_input = [] \n    mask_input = []\n    seg_input = []\n    \n    for text in input_text:\n        text = tokenizer.tokenize(text)\n        text = text[:max_len-2]\n        input_sequence = [\"[CLS]\"] + text + [\"[SEP]\"]\n        pad_len = max_len - len(input_sequence)\n        \n        tokens = tokenizer.convert_tokens_to_ids(input_sequence)      \n        token_input.append(tokens + [0]*pad_len)\n        mask_input.append([1]*len(input_sequence) + [0]*pad_len)\n        seg_input.append([0] * max_len)\n        \n    return np.array(token_input), np.array(mask_input), np.array(seg_input)","execution_count":65,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def build_model(bert_layer, max_len = 512):\n    input_word_ids = Input(shape=(max_len, ),dtype = tf.int32,name = 'input_words_ids')\n    input_mask = Input(shape=(max_len,), dtype=tf.int32, name=\"input_mask\")\n    segment_ids = Input(shape=(max_len,), dtype=tf.int32, name=\"segment_ids\")\n    \n    _, sequence_output = bert_layer([input_word_ids, input_mask, segment_ids])\n    clf_output = sequence_output[:, 0, :]\n    out = Dense(1, activation='sigmoid')(clf_output)\n    \n    model = Model(inputs=[input_word_ids, input_mask, segment_ids], outputs=out)\n    model.compile(Adam(lr=2e-6), loss='binary_crossentropy', metrics=['accuracy'])\n    \n    return model","execution_count":66,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# X_train_bert.user_review.str.len()","execution_count":67,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# mean_len = X_train_bert.user_review.str.len().mean()","execution_count":68,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# mean_len","execution_count":69,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# X_test_bert.user_review.str.len().mean()","execution_count":70,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# X_train_bert = pd.Series(X_train_bert)\n# X_test_bert =pd.Series(X_test_bert)","execution_count":71,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"type(X_train_bert)","execution_count":72,"outputs":[{"output_type":"execute_result","execution_count":72,"data":{"text/plain":"pandas.core.series.Series"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"train_input = bert_encode(X_train_bert.values, tokenizer, max_len=400)\ntest_input = bert_encode(X_test_bert.values, tokenizer, max_len=400)\ntrain_labels = y_train.values","execution_count":73,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"testing_input = bert_encode(testing_df.user_review, tokenizer, max_len=400)","execution_count":74,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"model = build_model(bert_layer, max_len=400)\nmodel.summary()","execution_count":75,"outputs":[{"output_type":"stream","text":"Model: \"model\"\n__________________________________________________________________________________________________\nLayer (type)                    Output Shape         Param #     Connected to                     \n==================================================================================================\ninput_words_ids (InputLayer)    [(None, 400)]        0                                            \n__________________________________________________________________________________________________\ninput_mask (InputLayer)         [(None, 400)]        0                                            \n__________________________________________________________________________________________________\nsegment_ids (InputLayer)        [(None, 400)]        0                                            \n__________________________________________________________________________________________________\nkeras_layer (KerasLayer)        [(None, 1024), (None 335141889   input_words_ids[0][0]            \n                                                                 input_mask[0][0]                 \n                                                                 segment_ids[0][0]                \n__________________________________________________________________________________________________\ntf_op_layer_strided_slice (Tens [(None, 1024)]       0           keras_layer[0][1]                \n__________________________________________________________________________________________________\ndense (Dense)                   (None, 1)            1025        tf_op_layer_strided_slice[0][0]  \n==================================================================================================\nTotal params: 335,142,914\nTrainable params: 335,142,913\nNon-trainable params: 1\n__________________________________________________________________________________________________\n","name":"stdout"}]},{"metadata":{"trusted":true},"cell_type":"code","source":"train_history = model.fit(\n    train_input, train_labels,\n    validation_split=0.1,\n    epochs=3,\n    batch_size=5\n)","execution_count":76,"outputs":[{"output_type":"stream","text":"Train on 11808 samples, validate on 1312 samples\nEpoch 1/3\n11808/11808 [==============================] - 2131s 180ms/sample - loss: 0.3120 - accuracy: 0.8637 - val_loss: 0.2285 - val_accuracy: 0.9116\nEpoch 2/3\n11808/11808 [==============================] - 2086s 177ms/sample - loss: 0.1574 - accuracy: 0.9413 - val_loss: 0.2245 - val_accuracy: 0.9154\nEpoch 3/3\n11808/11808 [==============================] - 2085s 177ms/sample - loss: 0.0495 - accuracy: 0.9848 - val_loss: 0.2933 - val_accuracy: 0.9116\n","name":"stdout"}]},{"metadata":{"trusted":true},"cell_type":"code","source":"prediction = model.predict(test_input)\npreds = []\nfor x in prediction:\n    preds.append(int(x.round()))\n\nfrom sklearn.metrics import accuracy_score\nprint(\"Accuracy: \", accuracy_score(y_test.values,preds))","execution_count":77,"outputs":[{"output_type":"stream","text":"Accuracy:  0.9108367626886146\n","name":"stdout"}]},{"metadata":{"trusted":true},"cell_type":"code","source":"from sklearn.metrics import confusion_matrix \n\nprint(confusion_matrix(y_test.values, preds))","execution_count":78,"outputs":[{"output_type":"stream","text":"[[1626  216]\n [ 174 2358]]\n","name":"stdout"}]},{"metadata":{"trusted":true},"cell_type":"code","source":"preds.count(1)","execution_count":79,"outputs":[{"output_type":"execute_result","execution_count":79,"data":{"text/plain":"2574"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"preds.count(0)","execution_count":80,"outputs":[{"output_type":"execute_result","execution_count":80,"data":{"text/plain":"1800"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"labels = [0, 1]\ncm = confusion_matrix(y_test.values, preds, labels)\nprint(cm)\nfig = plt.figure()\nax = fig.add_subplot(111)\ncax = ax.matshow(cm)\nplt.title('Confusion matrix of the classifier')\nfig.colorbar(cax)\nax.set_xticklabels([''] + labels)\nax.set_yticklabels([''] + labels)\nplt.xlabel('Predicted')\nplt.ylabel('True')\nplt.show()","execution_count":81,"outputs":[{"output_type":"stream","text":"[[1626  216]\n [ 174 2358]]\n","name":"stdout"},{"output_type":"display_data","data":{"text/plain":"<Figure size 432x288 with 2 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"from sklearn.metrics import classification_report \nprint(classification_report(y_test.values, preds))","execution_count":82,"outputs":[{"output_type":"stream","text":"              precision    recall  f1-score   support\n\n           0       0.90      0.88      0.89      1842\n           1       0.92      0.93      0.92      2532\n\n    accuracy                           0.91      4374\n   macro avg       0.91      0.91      0.91      4374\nweighted avg       0.91      0.91      0.91      4374\n\n","name":"stdout"}]},{"metadata":{"trusted":true},"cell_type":"code","source":"# train_history = model.fit(\n#     train_input, train_labels,\n#     validation_split=0.1,\n#     epochs=1,\n#     batch_size=5\n# )","execution_count":83,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# prediction = model.predict(test_input)\n# preds = []\n# for x in prediction:\n#     preds.append(int(x.round()))\n\n# from sklearn.metrics import accuracy_score\n# print(\"Accuracy: \", accuracy_score(y_test.values,preds))","execution_count":84,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"pred_result = prediction = model.predict(testing_input)\nsubmission = pred_result.round().astype(int)\nsubmission=pd.DataFrame(submission)\n\nsub = pd.read_csv('/kaggle/input/steam-recommendation-nlp-dataset/test.csv')\nsubmission['review_id']=sub['review_id']\n\nsubmission= submission[['review_id',0]]\nsubmission.rename(columns = {0:'user_suggestion'}, inplace = True) \n\nsubmission.head(4)\n\n","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"submission.to_csv('submission_400_clean.csv', index=False)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# Experiment"},{"metadata":{"trusted":true},"cell_type":"code","source":"# X","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# super_train = bert_encode(X.user_review.values, tokenizer, max_len=380)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# super_prediction = model.predict(super_train)\n# super_pred = []\n# for x in super_prediction:\n#     super_pred.append(int(x.round()))\n\n","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# len(super_pred)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# from sklearn.metrics import accuracy_score\n# print(\"Accuracy: \", accuracy_score(super_pred, y.values))","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# now adding sentiment to it"},{"metadata":{"trusted":true},"cell_type":"code","source":"# from tqdm import tqdm\n# super_sentiment = []\n# for review in tqdm(X.user_review.values):\n#     blob = TextBlob(review)\n#     super_sentiment += [blob.sentiment.polarity]","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# def num(x):\n#     if x >= 0:\n#         return 1\n#     elif x <= 0:\n#         return -1","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# super_dataset = pd.DataFrame(super_pred)\n# super_dataset['sentiment']=super_sentiment\n# super_dataset['Target']=y.values","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# super_dataset.tail(10)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# super_dataset['sentiment']=super_dataset.sentiment.apply(num)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# super_dataset.sample(10)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# super_dataset.shape","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# X_super= super_dataset.drop(['Target'],axis=1)\n# y_super = super_dataset.Target","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# from sklearn.model_selection import train_test_split\n\n# X_train_super, X_test_super, y_train_super, y_test_super = train_test_split(X_super, y_super, test_size = 0.30, random_state = 0)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# print(X_train_super.shape)\n# print(X_test_super.shape)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# from sklearn.preprocessing import StandardScaler\n\n# scaler = StandardScaler()\n\n# X_train_super = pd.DataFrame(scaler.fit_transform(X_train_super), columns = X_super.columns)\n\n# X_test_super = pd.DataFrame(scaler.transform(X_test_super), columns = X_super.columns)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# X_train_super.head()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# from sklearn.metrics import accuracy_score","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# from sklearn.linear_model import LogisticRegression\n\n# clf = LogisticRegression(max_iter=200, )\n# clf.fit(X_train_super, y_train_super)\n# y_pred_super = clf.predict(X_test_super)\n\n# print('LogisticRegression accuracy_score {0:0.4f}'. format(accuracy_score(y_test_super, y_pred_super)))\n# clf.score(X_train_super, y_train_super)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# from xgboost import XGBClassifier\n\n# clf= XGBClassifier(learning_rate=0.04, n_estimators=406, max_depth=5,\n#                         min_child_weight=4, \n#                          seed=27)\n# clf.fit(X_train_super, y_train_super)\n# y_pred_super = clf.predict(X_test_super)\n\n# print('XGBClassifier accuracy_score {0:0.4f}'. format(accuracy_score(y_test_super, y_pred_super)))\n# clf.score(X_train_super, y_train_super)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# from sklearn.tree import DecisionTreeClassifier\n\n# clf = DecisionTreeClassifier( max_depth=5)\n# clf.fit(X_train_super, y_train_super)\n# y_pred_super = clf.predict(X_test_super)\n\n# print('DecisionTreeClassifier accuracy_score {0:0.4f}'. format(accuracy_score(y_test_super, y_pred_super)))\n# clf.score(X_train_super, y_train_super)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# from catboost import CatBoostClassifier\n# clf= CatBoostClassifier( iterations=100,\n#                 learning_rate =0.05,\n#                  max_depth=4,\n#                  leaf_estimation_iterations=20)\n# clf.fit(X_train_super, y_train_super)\n# y_pred_super = clf.predict(X_test_super)\n\n# print('CatBoostClassifier accuracy_score {0:0.4f}'. format(accuracy_score(y_test_super, y_pred_super)))\n# clf.score(X_train_super, y_train_super)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# from sklearn.ensemble import GradientBoostingClassifier\n# clf= GradientBoostingClassifier(                          \n#     )\n# clf.fit(X_train_super, y_train_super)\n# y_pred_super = clf.predict(X_test_super)\n\n# print('GradientBoostingClassifier accuracy_score {0:0.4f}'. format(accuracy_score(y_test_super, y_pred_super)))\n# clf.score(X_train_super, y_train_super)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# from sklearn.ensemble import BaggingClassifier\n# clf= BaggingClassifier(#base_estimator=base_cls,\n#     n_estimators=400,\n#    )\n# clf.fit(X_train_super, y_train_super)\n# y_pred_super = clf.predict(X_test_super)\n\n# print('BaggingClassifier accuracy_score {0:0.4f}'. format(accuracy_score(y_test_super, y_pred_super)))\n# clf.score(X_train_super, y_train_super)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# from sklearn.experimental import enable_hist_gradient_boosting\n# from sklearn.ensemble import HistGradientBoostingClassifier\n\n# clf= HistGradientBoostingClassifier( #loss='binary_crossentropy',\n#     learning_rate=0.05,\n# #     max_iter=6000,\n#     max_depth=6,\n# #     max_leaf_nodes=5,\n#    )\n# clf.fit(X_train_super, y_train_super)\n# y_pred_super = clf.predict(X_test_super)\n\n# print('HistGradientBoostingClassifier accuracy_score {0:0.4f}'. format(accuracy_score(y_test_super, y_pred_super)))\n# clf.score(X_train_super, y_train_super)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# from lightgbm import LGBMClassifier\n\n# clf= LGBMClassifier(# bagging_freq = 5,\n#     bagging_fraction = 1,\n#     boost_from_average= 'true',\n#     boost =  'gbdt',\n# #     feature_fraction =  0.01,\n#     learning_rate =  0.05,\n# #     max_depth =  15,  \n# #     metric = log_loss,\n#     min_data_in_leaf =  10,\n#     min_sum_hessian_in_leaf =  0.05,\n#     num_leaves =  14,\n# #     num_threads =  8,\n#     tree_learner =  'serial',\n#     verbosity =  1\n#                    )\n# clf.fit(X_train_super, y_train_super)\n# y_pred_super = clf.predict(X_test_super)\n\n# print('LGBMClassifier accuracy_score {0:0.4f}'. format(accuracy_score(y_test_super, y_pred_super)))\n# clf.score(X_train_super, y_train_super)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# Testing DataFrame Super_output"},{"metadata":{"trusted":true},"cell_type":"code","source":"# submission_super=pd.DataFrame(submission)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# submission_super.drop(\"review_id\",axis=1,inplace=True)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# submission_super.head()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# submission_super.shape","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# from tqdm import tqdm\n# submission_super_sentiment = []\n# for review in tqdm(testing_df.user_review.values):\n#     blob = TextBlob(review)\n#     submission_super_sentiment += [blob.sentiment.polarity]","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# len(submission_super_sentiment)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# submission_super['sentiment']=submission_super_sentiment","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# submission_super.head()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# submission_super['sentiment']=submission_super.sentiment.apply(num)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# submission_super = pd.DataFrame(scaler.transform(submission_super), columns = submission_super.columns)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# submission_super.rename(columns = {'user_suggestion':0}, inplace = True) \n# submission_super.head()","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# saving the super submission"},{"metadata":{"trusted":true},"cell_type":"code","source":"# X_test_super","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# super_predictions = clf.predict(submission_super)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# sss = pd.read_csv('/kaggle/input/steam-recommendation-nlp-dataset/test.csv')","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# result_data_df = pd.DataFrame(sss.review_id)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# result_data_df['user_suggestion']=super_predictions","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# result_data_df.head()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# result_data_df.to_csv('exp21.csv', index=False)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# Experiment End"},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# Saving Output by BerT"},{"metadata":{"trusted":true},"cell_type":"code","source":"# pred_result = prediction = model.predict(testing_input)\n# submission = pred_result.round().astype(int)\n# submission=pd.DataFrame(submission)\n\n# sub = pd.read_csv('/kaggle/input/steam-recommendation-nlp-dataset/test.csv')\n# submission['review_id']=sub['review_id']\n\n# submission= submission[['review_id',0]]\n# submission.rename(columns = {0:'user_suggestion'}, inplace = True) \n\n# submission.head(4)\n\n# submission.to_csv('submission_91.csv', index=False)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"\n","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]}],"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat":4,"nbformat_minor":4}