{"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_minor":4,"nbformat":4,"cells":[{"cell_type":"code","source":"# This Python 3 environment comes with many helpful analytics libraries installed\n# It is defined by the kaggle/python Docker image: https://github.com/kaggle/docker-python\n# For example, here's several helpful packages to load\nimport os\nimport time\nimport numpy as np # linear algebra\nimport pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\nimport re\nimport warnings\nwarnings.filterwarnings(\"ignore\")\n# sns to plot data\nimport seaborn as sns\nimport matplotlib.pyplot as plt\n# Wordcloud to visualize words\nfrom wordcloud import WordCloud, STOPWORDS, ImageColorGenerator\n\nimport nltk\nfrom nltk.corpus import stopwords \nfrom nltk.tokenize import word_tokenize \n# Import stopwords\nstopwords = stopword_list = nltk.corpus.stopwords.words('english')\n\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.feature_extraction.text import CountVectorizer\nfrom sklearn.metrics import accuracy_score, f1_score\nfrom sklearn.linear_model import LogisticRegression","metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","execution":{"iopub.status.busy":"2021-06-11T00:35:46.240189Z","iopub.execute_input":"2021-06-11T00:35:46.240733Z","iopub.status.idle":"2021-06-11T00:35:47.978452Z","shell.execute_reply.started":"2021-06-11T00:35:46.240650Z","shell.execute_reply":"2021-06-11T00:35:47.977295Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# 1. Phân tích dữ liệu\n\nMục đích của bước này: đọc và phân tích các đặc trưng của 2 bộ dữ liệu: câu hỏi sincere (tạm dịch là câu hỏi chân thành) và câu hỏi insincere (tạm dịch là câu hỏi không chân thành), cũng như làm các bước cẩn thiết để xử lý dữ liệu trước khi đưa vào mô hình huấn luyện, sao cho mô hình đạt hiệu năng tốt nhất.\n\n*(Ngoài ra, đa số các đoạn code đều có comment bên trên để nói qua mục đích và kết quả dự kiến của dòng code đó.)*","metadata":{}},{"cell_type":"markdown","source":"#### Đoạn code dưới có mục đích đơn giản là nhập 2 bộ dữ liệu được Quora cho sẵn: train (dữ liệu dùng để huấn luyện) và test (dữ liệu dùng để đánh giá hiệu năng của mô hình)","metadata":{}},{"cell_type":"code","source":"# Data import\ntrain_data = pd.read_csv(\"../input/quora-insincere-questions-classification/train.csv\")\ntest_data = pd.read_csv(\"../input/quora-insincere-questions-classification/test.csv\")\n\n# Print imported datapoints\nprint(f\"{len(train_data):,} datapoints\")\nprint(f\"{len(test_data):,} test datapoints\")\n\n# Print all available columns in csv\nprint(train_data.head())","metadata":{"execution":{"iopub.status.busy":"2021-06-11T00:35:47.980117Z","iopub.execute_input":"2021-06-11T00:35:47.980408Z","iopub.status.idle":"2021-06-11T00:35:53.620813Z","shell.execute_reply.started":"2021-06-11T00:35:47.980381Z","shell.execute_reply":"2021-06-11T00:35:53.619824Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"##### Như kết quả phía trên: ta thu được hơn 1.3 triệu câu hỏi dùng để train, và gần 400 nghìn câu hỏi dùng để đánh giá hiệu năng của mô hình. Ngay bên dưới sẽ là format dữ liệu của bộ ```train_data```, bao gồm:\n\n* Cột ```qid```: Đây chính là id của câu hỏi, mỗi câu hỏi đều có 1 id khác nhau và không có id nào giống nhau cả.\n* Cột ```question_text```: Đây chính là cột chứa các câu hỏi. Chúng ta sẽ phân tích và sàng lọc cột này trước khi đưa vào mô hình để huấn luyện, đảm bảo dữ liệu sẽ \"sạch\" nhất có thể.\n* Cột ```target```: Đây chính là cột chứa kết quả đánh giá câu hỏi có phải chân thành hay không chân thành, câu hỏi ```question_text``` có ```target = 0``` sẽ được đánh giá là câu hỏi chân thành, và câu hỏi có ```target = 1``` sẽ được đánh giá là không chân thành.","metadata":{}},{"cell_type":"markdown","source":"#### Đoạn code tiếp theo sẽ dùng để biểu diễn dữ liệu ```train_data``` thành một đồ thị, trong đó chiều ngang của đồ thị sẽ biểu diễn loại câu hỏi, còn chiều dọc sẽ biểu thị tổng số câu hỏi có trong loại câu hỏi đó.","metadata":{}},{"cell_type":"code","source":"sns.countplot(train_data['target'])","metadata":{"execution":{"iopub.status.busy":"2021-06-11T00:35:53.622384Z","iopub.execute_input":"2021-06-11T00:35:53.622655Z","iopub.status.idle":"2021-06-11T00:35:53.874901Z","shell.execute_reply.started":"2021-06-11T00:35:53.622629Z","shell.execute_reply":"2021-06-11T00:35:53.873676Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#### Nhật xét sơ bộ về phân bố dữ liệu: Dữ liệu bên trên bao gồm có khoảng 1.2 triệu câu hỏi chân thành, và khoảng 100 nghìn câu hỏi không chân thành. Số lượng câu hỏi chân thành chiếm tới hơn 93% tổng số bộ dữ liệu, vậy nên dữ liệu dùng để huấn luyện rất không cân bằng, cần phải đề ra giải pháp để cân bằng lại dữ liệu.","metadata":{}},{"cell_type":"markdown","source":"## Quá trình phân tích và sàng lọc dữ liệu","metadata":{}},{"cell_type":"markdown","source":"#### Đoạn code bên dưới sẽ kiểm tra xem có tập dữ liệu nào trong tập ```train_data``` và ```test_data``` là ```null``` hay không\n","metadata":{}},{"cell_type":"code","source":"print('train_data null datapoints: ', train_data.isnull().sum())\ntest_data.isnull().sum()","metadata":{"execution":{"iopub.status.busy":"2021-06-11T00:35:53.876456Z","iopub.execute_input":"2021-06-11T00:35:53.876800Z","iopub.status.idle":"2021-06-11T00:35:54.024727Z","shell.execute_reply.started":"2021-06-11T00:35:53.876751Z","shell.execute_reply":"2021-06-11T00:35:54.023835Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#### Đoạn code bên dưới sẽ vẽ ra biểu đồ box plot để xác định phân bố độ dài của các câu hỏi trong từng loại câu hỏi khác nhau. Cách đọc đồ thị box plot sẽ được hiển thị bên dưới:\n![Cách đọc đồ thị box plot](https://www.simplypsychology.org/boxplot.jpg)\n#### Giải thích sơ bộ:\n* ```min```, ```max```: Đây chính là 2 giá trị nhỏ nhất và lớn nhất của tập dữ liệu trong box plot.\n* Lower quartile ```Q1```: 25% giá trị trong tập dữ liệu thấp hơn đường ```Q1``` sẽ nằm dưới đường kẻ đó.\n* Median: 50% giá trị trong tập dữ liệu thấp hơn đường median sẽ nằm dưới đường kẻ đó, 50% còn lại sẽ nằm bên trên.\n* Upper quartile ```Q2```: 75% giá trị trong tập dữ liệu thấp hơn đường ```Q2``` sẽ nằm dưới đường kẻ đó, 25% còn lại sẽ nằm bên trên.\n* Khung interquartile ```IQR```: khoảng dữ liệu nằm giữa 25% và 75% của dữ liệu sẽ nằm trong khoảng này. Các thông số bên trong ```IQR``` đã được mô tả bên trên.","metadata":{}},{"cell_type":"code","source":"# Get question length and then plot\nquestion_length = []\nfor x in (train_data['question_text']):\n    question_length.append(len(x))\n# np.append(train_data, question_length, axis=1)\ntrain_data['question_length'] = question_length\nsns.boxplot(x=\"target\", y='question_length', data=train_data)\ntrain_data.drop('question_length', axis=1)","metadata":{"execution":{"iopub.status.busy":"2021-06-11T00:35:54.025846Z","iopub.execute_input":"2021-06-11T00:35:54.026212Z","iopub.status.idle":"2021-06-11T00:35:55.023720Z","shell.execute_reply.started":"2021-06-11T00:35:54.026178Z","shell.execute_reply":"2021-06-11T00:35:55.022791Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#### Nhận xét phân bố độ dài các câu hỏi trong tập dữ liệu huấn luyện:\n* Độ dài các câu hỏi từ các tập câu hỏi chân thành luôn nhỏ hơn độ dài câu hỏi trong tập không chân thành: trừ điểm ```min``` ra thì Q1, Q2, median của câu hỏi chân thành luôn có giá trị thấp hơn không chân thành. Từ đây ta có 2 phỏng đoán:\n    * Đa số các câu hỏi không chân thành đều dài hơn các câu hỏi chân thành (có thể là các câu hỏi spam, các câu hỏi mang tính chất quảng cáo, chửi bới, ...)\n    * Bộ dữ liệu câu hỏi không chân thành bị chuẩn đoán sai bởi các câu hỏi có quá nhiều độ nhiễu (ví dụ như có quá nhiều ký tự đặc biệt, ký tự HTML, ký tự toán học, ...). Vậy nên có thể có câu hỏi toàn chứa các ký tự HTML (như ```<div></div>```, ```<body></body>```, ...) đều bị đánh dấu là không chân thành, mặc dù nó không hề như vậy.\n* Câu hỏi có độ dài lớn nhất trong tập câu hỏi chân thành rơi vào khoảng 800 ký tự. Ta sẽ xem rõ câu hỏi này là gì.\n* Câu hỏi có độ dài lớn nhất trong tập câu hỏi không chân thành có thể lên tới 1000 ký tự, và cũng là câu hỏi có độ dài lớn nhất trong tập dữ liệu huấn luyện. Ta cũng sẽ phải xem xét câu hỏi này.\n\n#### Ngoài ra, ta sẽ bỏ cột ```qid``` trong tập dữ liệu vì không cần thiết, và để giải phóng bộ nhớ vì quá trình huấn luyện sẽ cần rất nhiều bộ nhớ.","metadata":{}},{"cell_type":"markdown","source":"#### Đoạn code dưới sẽ tìm độ dài lớn nhất và độ dài lớn nhì của 2 loại câu hỏi trong tập dữ liệu huấn luyện\n\n*(Trường ```question_length``` đã được thêm vào ở trên đoạn code vẽ box plot.)*","metadata":{}},{"cell_type":"code","source":"# Find max for both sincere and insincere question length\nmax_sincere = 0\nsecond_max_sincere = 0\nmax_insincere = 0\nsecond_max_insincere = 0\n\nfor x in train_data.itertuples():\n#     print(x[4])\n    if (x[3] == 0 and x[4] > max_sincere):\n        max_sincere = x[4]\n    if (x[3] == 1 and x[4] > max_insincere):\n        max_insincere = x[4]\n\nfor x in train_data.itertuples():\n    if (x[3] == 0 and x[4] < max_sincere and x[4] > second_max_sincere):\n        second_max_sincere = x[4]\n    if (x[3] == 1 and x[4] < max_insincere and x[4] > second_max_insincere):\n        second_max_insincere = x[4]\n        \nprint('Max sincere question length: ', max_sincere)\nprint('Max insincere question length: ', max_insincere)\n\nprint('Second max sincere question length: ', second_max_sincere)\nprint('Second max insincere question length: ', second_max_insincere)\n","metadata":{"execution":{"iopub.status.busy":"2021-06-11T00:35:55.025123Z","iopub.execute_input":"2021-06-11T00:35:55.025415Z","iopub.status.idle":"2021-06-11T00:35:57.369591Z","shell.execute_reply.started":"2021-06-11T00:35:55.025388Z","shell.execute_reply":"2021-06-11T00:35:57.368603Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Find the question with that length\nfor x in train_data.itertuples():\n    if (x[4] == max_sincere and x[3] == 0):\n        print('Sincere question: ', x[2], '\\n')\n    if (x[4] == max_insincere and x[3] == 1):\n        print('Insincere question: ', x[2], '\\n')\n        \nprint('\\n\\n\\n')\n        \nfor x in train_data.itertuples():\n    if (x[4] == second_max_sincere and x[3] == 0):\n        print('Second max sincere question: ', x[2], '\\n')\n    if (x[4] == second_max_insincere and x[3] == 1):\n        print('Second max insincere question: ', x[2], '\\n')","metadata":{"execution":{"iopub.status.busy":"2021-06-11T00:35:57.370855Z","iopub.execute_input":"2021-06-11T00:35:57.371386Z","iopub.status.idle":"2021-06-11T00:35:59.568628Z","shell.execute_reply.started":"2021-06-11T00:35:57.371345Z","shell.execute_reply":"2021-06-11T00:35:59.567643Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#### Nhận xét kết quả thu được bên trên:\n* Ở các câu hỏi có độ dài lớn nhất của 2 loại câu hỏi, ta có thể thấy rõ rằng câu hỏi không chân thành chính là một câu hỏi toán học, và có rất nhiều ký tự đặc biệt trong đó, nhưng điều lạ là câu hỏi này bị đánh dấu là không chân thành, mặc dù nó chỉ là một câu hỏi toán học thông thường? Nhìn qua ta biết được đây chính là các ký hiệu được dùng bởi LaTeX để viết ra những công thức toán học. Dùng công cụ LaTeX Viewer có sẵn trên mạng, ta thu được công thức toán học sau: \n![image.png](attachment:bdabcd5e-1aff-4abf-8171-13a6b55fc65e.png)\n    * Ta thấy rõ là câu hỏi trên không có gì đáng nghi ngờ về mặt ngữ nghĩa, tuy nhiên, vì khi chuyển thể từ dạng MathML về văn bản thông thường, các ký hiệu đánh dấu vẫn được giữ nguyên, làm nhiễu dữ liệu, vậy nên bị đánh dấu không chính xác.\n* Câu hỏi chân thành có độ dài lớn nhất đã được đánh dấu đúng, không có gì đáng nghi ở đây.\n* Ở các câu hỏi có độ dài lớn nhì của 2 loại câu hỏi, ta thấy rằng cả 2 câu hỏi đều có rất nhiều ký hiệu toán học làm nhiễu dữ liệu. Hình ảnh LaTeX của 2 câu hỏi trên được hiển thị như dưới:\n    ![image.png](attachment:a0a86dc0-d687-4406-98c4-4f05c1001554.png)\n\n    *Câu hỏi không chân thành có độ dài lớn nhì*\n    \n    ![image.png](attachment:9ed77e08-e600-4433-bfda-1fc431d3971d.png)\n    *Câu hỏi chân thành có độ dài lớn nhì*\n* Một lần nữa, mặt ngữ nghĩa của 2 câu hỏi không có gì đáng ngờ, nhưng số ký tự trong câu hỏi không chân thành (878) lớn hơn nhiều so với câu hỏi chân thành (509), vậy nên có khả năng là nó bị đánh dấu không chân thành vì có quá nhiều ký tự đặc biệt\n\n#### Ta phải loại bỏ các ký tự không thuộc trong bảng chữ cái, và có thể sẽ loại bỏ chữ số, nhưng vì lý do sẽ nêu bên dưới khiến em sẽ không loại bỏ số mà chỉ loại bỏ các ký tự không thuộc bảng chữ cái thôi.","metadata":{},"attachments":{"bdabcd5e-1aff-4abf-8171-13a6b55fc65e.png":{"image/png":"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"},"a0a86dc0-d687-4406-98c4-4f05c1001554.png":{"image/png":"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"},"9ed77e08-e600-4433-bfda-1fc431d3971d.png":{"image/png":"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"}}},{"cell_type":"markdown","source":"#### Đoạn code bên dưới sẽ sử dụng thư viện WordCloud để hình dung hoá tần suất xuất hiện của các từ có trong câu hỏi.\n","metadata":{}},{"cell_type":"code","source":"# WordCloud function\ndef cloud(text, title, size = (10,7)):\n    # Process text\n    wordcloud = WordCloud(width=800, height=400, collocations=False).generate(\" \".join(text))\n    \n    # Visualization\n    fig = plt.figure(figsize=size, dpi=80, )\n    plt.imshow(wordcloud,interpolation='bilinear')\n    # Show title but not working?\n    plt.title(title, fontsize=25,color='w')\n    plt.tight_layout(pad=1)\n    plt.show()","metadata":{"execution":{"iopub.status.busy":"2021-06-11T00:35:59.571389Z","iopub.execute_input":"2021-06-11T00:35:59.571832Z","iopub.status.idle":"2021-06-11T00:35:59.577514Z","shell.execute_reply.started":"2021-06-11T00:35:59.571768Z","shell.execute_reply":"2021-06-11T00:35:59.576393Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Separate sincere and insincere datapoints for visualization\n\ndef separate_data(data, target):\n    new_data = []\n    # Itertuples to iterate each row in train_data\n    for x in data.itertuples():\n        # index 3 represents 'target', 2 represents question_text\n        if (x[3] == target): new_data.append(x[2])\n    return new_data\n\nsincere_train_data = separate_data(train_data, 0)\ninsincere_train_data = separate_data(train_data, 1)","metadata":{"execution":{"iopub.status.busy":"2021-06-11T00:35:59.579146Z","iopub.execute_input":"2021-06-11T00:35:59.579424Z","iopub.status.idle":"2021-06-11T00:36:01.640106Z","shell.execute_reply.started":"2021-06-11T00:35:59.579398Z","shell.execute_reply":"2021-06-11T00:36:01.639201Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#### Tần suất các từ có trong tập câu hỏi không chân thành","metadata":{}},{"cell_type":"code","source":"cloud(insincere_train_data, 'Insincere questions')","metadata":{"execution":{"iopub.status.busy":"2021-06-11T00:36:01.641186Z","iopub.execute_input":"2021-06-11T00:36:01.641438Z","iopub.status.idle":"2021-06-11T00:36:04.431019Z","shell.execute_reply.started":"2021-06-11T00:36:01.641414Z","shell.execute_reply":"2021-06-11T00:36:04.429959Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#### Tần suất các từ có trong tập câu hỏi không chân thành","metadata":{}},{"cell_type":"code","source":"cloud(sincere_train_data, 'Sincere questions')","metadata":{"execution":{"iopub.status.busy":"2021-06-11T00:36:04.432598Z","iopub.execute_input":"2021-06-11T00:36:04.433028Z","iopub.status.idle":"2021-06-11T00:36:22.760936Z","shell.execute_reply.started":"2021-06-11T00:36:04.432986Z","shell.execute_reply":"2021-06-11T00:36:22.759926Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#### Nhận xét về tần suất các từ trong tập huấn luyện: \n* Trong các từ xuất hiện nhiều nhất (biểu diễn bởi chữ to: chữ càng to thì tần suất xuất hiện càng nhiều), có một chút sự tương đồng giữa các từ (ví dụ như people, will, ...).\n* Tuy nhiên, tần suất các từ stop word (như những động từ to-be: is, are, am, ...) có tần suất xuất hiện rất nhiều, có thể ảnh hưởng tới quá trình học, nên ta phải loại bỏ các từ đó sử dụng một tập từ điển stop word có sẵn\n","metadata":{}},{"cell_type":"markdown","source":"# 2. Resampling\n#### Bởi dữ liệu trong tập ```train_data``` không cân bằng, nên ta cần phải tăng lượng câu hỏi không chân thành lên sao cho tỉ lệ câu hỏi chân thành và câu hỏi không chân thành đạt tỉ lệ chấp nhận được.\n\n#### Đoạn code bên dưới sẽ phục vụ mục đích trên, sử dụng hàm ```resample``` có trong ```sklearn```","metadata":{}},{"cell_type":"code","source":"from sklearn.utils import resample\n\nsincere_questions = train_data[train_data.target == 0]\ninsincere_questions = train_data[train_data.target == 1]\n\n# # 1:3 Ratio\n# new_train_data = pd.concat([resample(sincere_questions, replace = True, n_samples = len(insincere_questions)*3), insincere_questions])\n\nnew_train_data = pd.concat([resample(sincere_questions, replace = True, n_samples = len(insincere_questions)*5), insincere_questions])","metadata":{"execution":{"iopub.status.busy":"2021-06-11T00:36:22.762125Z","iopub.execute_input":"2021-06-11T00:36:22.762392Z","iopub.status.idle":"2021-06-11T00:36:23.138082Z","shell.execute_reply.started":"2021-06-11T00:36:22.762366Z","shell.execute_reply":"2021-06-11T00:36:23.137283Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"sns.countplot(new_train_data['target'])","metadata":{"execution":{"iopub.status.busy":"2021-06-11T00:36:23.139212Z","iopub.execute_input":"2021-06-11T00:36:23.139845Z","iopub.status.idle":"2021-06-11T00:36:23.337794Z","shell.execute_reply.started":"2021-06-11T00:36:23.139778Z","shell.execute_reply":"2021-06-11T00:36:23.336933Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Set new_train_data to train_data\ntrain_data = new_train_data","metadata":{"execution":{"iopub.status.busy":"2021-06-11T00:36:23.338810Z","iopub.execute_input":"2021-06-11T00:36:23.339054Z","iopub.status.idle":"2021-06-11T00:36:23.367526Z","shell.execute_reply.started":"2021-06-11T00:36:23.339030Z","shell.execute_reply":"2021-06-11T00:36:23.366760Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#### Đoạn code bên dưới chỉ phục vụ mục đích là chuyển cột ```target``` sang một mảng mới.","metadata":{}},{"cell_type":"code","source":"# Separate data labels\ntarget_array = train_data.pop('target')\ntarget_array.value_counts(True)\n\n#Drop train_data qid as it is unneccesary\ntrain_data.drop('qid', axis=1, inplace=True)\n","metadata":{"execution":{"iopub.status.busy":"2021-06-11T00:36:23.368525Z","iopub.execute_input":"2021-06-11T00:36:23.368974Z","iopub.status.idle":"2021-06-11T00:36:23.427031Z","shell.execute_reply.started":"2021-06-11T00:36:23.368932Z","shell.execute_reply":"2021-06-11T00:36:23.426111Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# 3. Tiền xử lý dữ liệu\n\n#### Đoạn code bên dưới sẽ có mục đích như sau:\n* Loại bỏ các ký tự không có trong bảng chữ cái (trừ dấu cách), bằng cách dùng regular expression: hiệu năng nhanh hơn là tìm và thay thế\n* Trước khi loại bỏ các ký tự, tìm các từ viết tắt (như it's, aren't, doesn't, ...) và khôi phục về nguyên bản (it is, are not, does not, ...)\n* KHÔNG loại bỏ số (có các số như 69, 420 là những số có ngữ nghĩa liên quan tới mức độ chân thành và không chân thành)\n#### Tất cả việc tìm và thay thế sẽ sử dụng ```regex``` vì có hiệu năng nhanh hơn.\n","metadata":{}},{"cell_type":"code","source":"#Preprocess\n\n#Remove all non-characters\nnon_char_regex= re.compile(r'[^A-Za-z0-9]+')\n# Separate sentence into array of words\nfrom nltk.tokenize import word_tokenize\nfrom nltk.stem import WordNetLemmatizer\n# Group similar words and treat them as one word.\nl = WordNetLemmatizer()\n\n# Common miss-spell dictionary, taken from various online sites\nmiss_spell_dictionary = {\n  'colour': 'color',\n  'centre': 'center',\n  'favourite': 'favorite',\n  'travelling': 'traveling',\n  'counselling': 'counseling',\n  'theatre': 'theater',\n  'cancelled': 'canceled',\n  'labour': 'labor',\n  'organisation': 'organization',\n  'wwii': 'world war 2',\n  'citicise': 'criticize',\n  'youtu ': 'youtube ',\n  'Qoura': 'Quora',\n  'sallary': 'salary',\n  'Whta': 'What',\n  'narcisist': 'narcissist',\n  'howdo': 'how do',\n  'whatare': 'what are',\n  'howcan': 'how can',\n  'howmuch': 'how much',\n  'howmany': 'how many',\n  'whydo': 'why do',\n  'doI': 'do I',\n  'theBest': 'the best',\n  'howdoes': 'how does',\n  'mastrubation': 'masturbation',\n  'mastrubate': 'masturbate',\n  \"mastrubating\": 'masturbating',\n  'pennis': 'penis',\n  'Etherium': 'bitcoin',\n  'narcissit': 'narcissist',\n  'bigdata': 'big data',\n  '2k17': '2017',\n  '2k18': '2018',\n  'qouta': 'quota',\n  'exboyfriend': 'ex boyfriend',\n  'airhostess': 'air hostess',\n  \"whst\": 'what',\n  'watsapp': 'whatsapp',\n  'demonitisation': 'demonetization',\n  'demonitization': 'demonetization',\n  'demonetisation': 'demonetization',\n  'electroneum': 'bitcoin',\n  'nanodegree': 'degree',\n  'hotstar': 'star',\n  'dream11': 'dream',\n  'ftre': 'fire',\n  'tensorflow': 'framework',\n  'unocoin': 'bitcoin',\n  'lnmiit': 'limit',\n  'unacademy': 'academy',\n  'altcoin': 'bitcoin',\n  'altcoins': 'bitcoin',\n  'litecoin': 'bitcoin',\n  'coinbase': 'bitcoin',\n  'cryptocurency': 'cryptocurrency',\n  'simpliv': 'simple',\n  'quoras': 'quora',\n  'schizoids': 'psychopath',\n  'remainers': 'remainder',\n  'twinflame': 'soulmate',\n  'quorans': 'quora',\n  'brexit': 'demonetized',\n  'iiest': 'institute',\n  'dceu': 'comics',\n  'pessat': 'exam',\n  'uceed': 'college',\n  'bhakts': 'devotee',\n  'boruto': 'anime',\n  'cryptocoin': 'bitcoin',\n  'blockchains': 'blockchain',\n  'fiancee': 'fiance',\n  'redmi': 'smartphone',\n  'oneplus': 'smartphone',\n  'qoura': 'quora',\n  'deepmind': 'framework',\n  'ryzen': 'cpu',\n  'whattsapp': 'whatsapp',\n  'undertale': 'adventure',\n  'zenfone': 'smartphone',\n  'cryptocurencies': 'cryptocurrencies',\n  'koinex': 'bitcoin',\n  'zebpay': 'bitcoin',\n  'binance': 'bitcoin',\n  'whtsapp': 'whatsapp',\n  'reactjs': 'framework',\n  'bittrex': 'bitcoin',\n  'bitconnect': 'bitcoin',\n  'bitfinex': 'bitcoin',\n  'yourquote': 'your quote',\n  'whyis': 'why is',\n  'jiophone': 'smartphone',\n  'dogecoin': 'bitcoin',\n  'onecoin': 'bitcoin',\n  'poloniex': 'bitcoin',\n  '7700k': 'cpu',\n  'angular2': 'framework',\n  'segwit2x': 'bitcoin',\n  'hashflare': 'bitcoin',\n  '940mx': 'gpu',\n  'openai': 'framework',\n  'hashflare': 'bitcoin',\n  '1050ti': 'gpu',\n  'nearbuy': 'near buy',\n  'freebitco': 'bitcoin',\n  'antminer': 'bitcoin',\n  'filecoin': 'bitcoin',\n  'whatapp': 'whatsapp',\n  'empowr': 'empower',\n  '1080ti': 'gpu',\n  'crytocurrency': 'cryptocurrency',\n  '8700k': 'cpu',\n  'whatsaap': 'whatsapp',\n  'g4560': 'cpu',\n  'payymoney': 'pay money',\n  'fuckboys': 'fuck boys',\n  'intenship': 'internship',\n  'zcash': 'bitcoin',\n  'demonatisation': 'demonetization',\n  'narcicist': 'narcissist',\n  'mastuburation': 'masturbation',\n  'trignometric': 'trigonometric',\n  'cryptocurreny': 'cryptocurrency',\n  'howdid': 'how did',\n  'crytocurrencies': 'cryptocurrencies',\n  'phycopath': 'psychopath',\n  'bytecoin': 'bitcoin',\n  'possesiveness': 'possessiveness',\n  'scollege': 'college',\n  'humanties': 'humanities',\n  'altacoin': 'bitcoin',\n  'demonitised': 'demonetized',\n  'brasília': 'brazilia',\n  'accolite': 'accolyte',\n  'econimics': 'economics',\n  'varrier': 'warrier',\n  'quroa': 'quora',\n  'statergy': 'strategy',\n  'langague': 'language',\n  'splatoon': 'game',\n  '7600k': 'cpu',\n  'gate2018': 'gate 2018',\n  'in2018': 'in 2018',\n  'narcassist': 'narcissist',\n  'jiocoin': 'bitcoin',\n  'hnlu': 'hulu',\n  '7300hq': 'cpu',\n  'weatern': 'western',\n  'interledger': 'blockchain',\n  'deplation': 'deflation',\n  'cryptocurrencies': 'cryptocurrency',\n  'bitcoin': 'blockchain cryptocurrency',\n  \n}\n\n# Shortened, experimental contraction dict\ncontraction_dict = {\n    \"let's\": \"let us\",\n    \"won't\": \"will not\",\n    \"shan't\": \"shall not\",\n    \"'m\": \" am\",\n    \"'s\": \" is\",\n    \"'re\": \" are\",\n    \"'ve\": \" have\",\n    \"n't\": \" not\",\n    \"'d\": \" would\",\n    \"'ll\": \" will\",\n    \"in'\": \"ing\"    \n}\n\n#Restore miss-spelled words\nmiss_spell_regex = re.compile('(%s)' % '|'.join(miss_spell_dictionary.keys()))\ndef restore_miss_spell(text):\n    def replace(match):\n        return miss_spell_dictionary[match.group(0)]\n    return miss_spell_regex.sub(replace, text)\n\n# Restore contraction\ncontraction_re = re.compile('(%s)' % '|'.join(contraction_dict.keys()))\ndef restore_contraction(text):\n    def replace(match):\n        return contraction_dict[match.group(0)]\n    return contraction_re.sub(replace, text)\n\ndef remove_stopwords (text):\n    word_tokens = word_tokenize(text) \n    filtered_sentence = [w for w in word_tokens if not w in stopwords] \n    filtered_sentence = [] \n    for w in word_tokens: \n        if w not in stopwords: \n            filtered_sentence.append(w)\n    filtered_sentence = ' '.join(filtered_sentence)\n    return filtered_sentence\n\n# Main process\ndef process(text): \n    text = text.lower()\n    text = restore_contraction(text)\n    text = restore_miss_spell(text)\n    text = non_char_regex.sub(' ', text)\n    text = remove_stopwords(text)\n    text = ' '.join([l.lemmatize(word) for word in word_tokenize(text)])\n    return text \n\nall_texts = np.array([process(x[0]) for x in train_data.values])\nprint(\"Questions process done\")","metadata":{"execution":{"iopub.status.busy":"2021-06-11T00:36:23.428461Z","iopub.execute_input":"2021-06-11T00:36:23.428858Z","iopub.status.idle":"2021-06-11T00:39:02.222561Z","shell.execute_reply.started":"2021-06-11T00:36:23.428818Z","shell.execute_reply":"2021-06-11T00:39:02.221610Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"example_sentence = \"\"\"This is a sample sentence, showing off the stop words filtration.\"\"\"\n\nprint(remove_stopwords(example_sentence)) \n  ","metadata":{"execution":{"iopub.status.busy":"2021-06-11T02:07:20.198017Z","iopub.execute_input":"2021-06-11T02:07:20.198345Z","iopub.status.idle":"2021-06-11T02:07:20.203698Z","shell.execute_reply.started":"2021-06-11T02:07:20.198311Z","shell.execute_reply":"2021-06-11T02:07:20.202372Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"print(all_texts[:10])","metadata":{"execution":{"iopub.status.busy":"2021-06-11T00:39:02.228967Z","iopub.execute_input":"2021-06-11T00:39:02.229231Z","iopub.status.idle":"2021-06-11T00:39:02.243828Z","shell.execute_reply.started":"2021-06-11T00:39:02.229206Z","shell.execute_reply":"2021-06-11T00:39:02.241623Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# 4. Quá trình huấn luyện model\n#### Đầu tiên là chia tập ```train_data``` và ```test_data```. Bộ dữ liệu ```train_data``` sẽ được chia ra như sau: 80% tập dữ liệu sẽ được dùng để huấn luyện, và 20% còn lại sẽ được dùng để test.","metadata":{}},{"cell_type":"code","source":"train_x, test_x, train_y, test_y = train_test_split(all_texts, target_array.values, test_size=0.2, random_state=0)","metadata":{"execution":{"iopub.status.busy":"2021-06-11T00:39:02.245151Z","iopub.execute_input":"2021-06-11T00:39:02.245482Z","iopub.status.idle":"2021-06-11T00:39:03.439596Z","shell.execute_reply.started":"2021-06-11T00:39:02.245453Z","shell.execute_reply":"2021-06-11T00:39:03.438673Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#### Đoạn code bên dưới sẽ sử dụng tập ```train_data``` và ```test_data``` đã tách để tạo ra các vector đếm.\n\nTa sẽ sử dụng ```Vectorizer``` để tách các từ khác nhau trong 1 câu hỏi và đếm tần suất xuất hiện của các từ đó. Làm vậy thì ngữ pháp của các câu hỏi sẽ được bỏ qua và không quan trọng trong việc xác định câu hỏi chân thành hay không chân thành.\n\nCác từ khác nhau xuất hiện trong câu hỏi, cũng như tần suất của chúng sẽ được gộp vào một Bag-of-Words (gọi tắt là bow). Sẽ có 2 loại BoW: ```bow_train (Full Bag-of-Words)``` sẽ dùng để chứa những câu có số từ được trích xuất trong các câu là 1, còn ```bow_train2 (Reduced Bag-of-Words)``` sẽ chứa những câu hỏi có số từ được trích xuất trong các câu là 2. Để minh hoạ rõ hơn ta sẽ lấy 1 ví dụ\n\n```This document is the first document.```\n    \n```And this is the second document```\n    \n* Với ```bow_train```, bộ từ vựng tổng hợp từ 2 câu trên là: ```[and, document, first, is, this, the, second]```, với tần suất các từ đó xuất hiện trong câu đầu tiên là: ```[0, 2, 1, 1, 1, 1, 0]``` và trong câu thứ 2: ```[1, 1, 0, 1, 1, 1, 1]```\n* Với ```bow_train2```, bộ từ vựng tổng hợp từ 2 câu trên là: ```['and this', 'document is', 'first document', 'is the', 'second document', 'this document', 'the first', 'the second', 'this is']```, với tần suất xuất hiện trong câu 1 là: ```[0, 1, 1, 1, 0, 1, 1, 0, 0]```, và câu 2: ```[1, 0, 0, 1, 1, 0, 0, 1, 1]```\n\nTrong khi tạo bộ ```bow_train2```, có 3 tham số được đề ra trong hàm ```CountVectorizer```:\n* ```min_df```: Khi tạo bộ từ vựng, những từ xuất hiện có tần suất thấp hơn ```min_df``` sẽ được loại bỏ. Thiết lập tham số này sẽ giúp loại bỏ những từ ngữ có tần suất quá ít (như những câu hỏi có xen lẫn vài từ có ngôn ngữ khác tiếng Anh), tránh ảnh hưởng tới quá trình học máy.\n* ```max_df```: Khi tạo bộ từ vựng, những từ xuất hiện có tần suất cao hơn ```max_df``` sẽ được loại bỏ. Thiếp lập tham số này sẽ giúp loại bỏ những từ ngữ được dùng quá thường xuyên (như những từ stop words), nhưng vì chúng ta đã loại bỏ các từ stop word rồi, nên tham số sẽ được đặt là ```0.999```\n* ```ngram_range```: Số từ được ```Vectorizer``` lọc ra và đẩy vào từ điển.\n\nNgoài ra, ```Vectorizer``` còn có kèm theo bộ lọc các từ stop words thông dụng của tiếng Anh, ta có thể sử dụng nó thay cho bộ lọc stop words bên trên đã làm, nhưng về kết quả thì sẽ có thể khác nhau.\n\n*(Bộ ```bow_train2``` sẽ dùng chủ yếu là để đánh giá và thử nghiệm)*","metadata":{}},{"cell_type":"code","source":"vectorizer = CountVectorizer()\nvectorizer2 = CountVectorizer(min_df = 0.0001, max_df = 0.999, ngram_range = (2,2)) \nbow_train = vectorizer.fit_transform(train_x) \nbow_train2 = vectorizer2.fit_transform(train_x) \nprint(bow_train.shape)\nprint(bow_train2.shape)\nbow_test = vectorizer.transform(test_x)\nbow_test2 = vectorizer2.transform(test_x)","metadata":{"execution":{"iopub.status.busy":"2021-06-11T00:39:03.441024Z","iopub.execute_input":"2021-06-11T00:39:03.441425Z","iopub.status.idle":"2021-06-11T00:39:16.767795Z","shell.execute_reply.started":"2021-06-11T00:39:03.441381Z","shell.execute_reply":"2021-06-11T00:39:16.766929Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#### 2 đoạn code bên dưới chỉ phục vụ mục đích là dùng 2 BoW đã xây dựng bên trên và bắt đầu huấn luyện, cũng như kiểm tra hiệu năng của mô hình.\n#### Ngoài ra, thuật toán Decision Tree cũng được thêm vào để phục vụ mục đích so sánh hiệu năng của thuật toán Logistic Regression\n#### Một vài đặc điểm của mô hình Logistic Regression:\n* Trong bài toán này, mức độ ```penalty``` được đặt ở mức L2 (Ridge Regression), bởi mức L1 (Lasso Regression) không được hỗ trợ bởi thư viện ```lbfgs```","metadata":{}},{"cell_type":"code","source":"# Logistic Regression (full BoW)\n\nprint(\"Full BoW Result\")\nlogistic = LogisticRegression(C=3.5) \nlogistic.fit(bow_train, train_y) \ntrain_predictions = logistic.predict(bow_train)\ntrain_acc = accuracy_score(train_y, train_predictions)\ntrain_f1 = f1_score(train_y, train_predictions) \nprint(f\"Train accuracy: {train_acc:.2%}, F1: {train_f1:.4f}\") \ntest_predictions = logistic.predict(bow_test)\ntest_acc = accuracy_score(test_y, test_predictions) \ntest_f1 = f1_score(test_y, test_predictions) \nprint(f\"Test accuracy:  {test_acc:.2%}, F1: {test_f1:.4f}\")","metadata":{"execution":{"iopub.status.busy":"2021-06-11T01:14:05.015136Z","iopub.execute_input":"2021-06-11T01:14:05.015635Z","iopub.status.idle":"2021-06-11T01:14:17.226736Z","shell.execute_reply.started":"2021-06-11T01:14:05.015592Z","shell.execute_reply":"2021-06-11T01:14:17.225971Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Logistic Regression (reduced BoW)\n\nprint(f\"Reduced BoW Result\")\nlogistic2 = LogisticRegression(C=3.5) \nlogistic2.fit(bow_train2, train_y) \ntrain_predictions = logistic2.predict(bow_train2)\ntrain_acc = accuracy_score(train_y, train_predictions)\ntrain_f1 = f1_score(train_y, train_predictions) \nprint(f\"Train accuracy: {train_acc:.4%}, F1: {train_f1:.4f}\") \ntest_predictions = logistic2.predict(bow_test2)\ntest_acc = accuracy_score(test_y, test_predictions) \ntest_f1 = f1_score(test_y, test_predictions) \nprint(f\"Test accuracy:  {test_acc:.4%}, F1: {test_f1:.4f}\")\n","metadata":{"execution":{"iopub.status.busy":"2021-06-11T01:08:42.115306Z","iopub.execute_input":"2021-06-11T01:08:42.115586Z","iopub.status.idle":"2021-06-11T01:08:46.530886Z","shell.execute_reply.started":"2021-06-11T01:08:42.115557Z","shell.execute_reply":"2021-06-11T01:08:46.529910Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Decision Tree (full BoW)\nfrom sklearn.tree import DecisionTreeClassifier\n\nprint(f\"Results of decision tree on full bag-of-words\")\ndt = DecisionTreeClassifier(max_depth=30) \ndt.fit(bow_train, train_y)\ntrain_predictions = dt.predict(bow_train)\ntrain_acc = accuracy_score(train_y, train_predictions) \ntrain_f1 = f1_score(train_y, train_predictions) \nprint(f\"Training accuracy: {train_acc:.2%}, F1: {train_f1:.4f}\") \ntest_predictions = dt.predict(bow_test)\ntest_acc = accuracy_score(test_y, test_predictions) \ntest_f1 = f1_score(test_y, test_predictions) \nprint(f\"Testing accuracy:  {test_acc:.2%}, F1: {test_f1:.4f}\")","metadata":{"execution":{"iopub.status.busy":"2021-06-11T00:39:34.252890Z","iopub.execute_input":"2021-06-11T00:39:34.253346Z","iopub.status.idle":"2021-06-11T00:39:59.004630Z","shell.execute_reply.started":"2021-06-11T00:39:34.253311Z","shell.execute_reply":"2021-06-11T00:39:59.003232Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# 5. Tổng hợp kết quả thu được","metadata":{}},{"cell_type":"markdown","source":"## Kết quả full BoW chưa lọc stop word, chưa có resampling\n\nTrain accuracy: 95.66%, F1: 0.5759\n\nTest accuracy:  95.36%, F1: 0.5428\n\n\n## Kết quả reduced BoW chưa lọc stop word, chưa có resampling\n\nTrain accuracy: 95.25%, F1: 0.5190\n\nTest accuracy:  95.16%, F1: 0.5082","metadata":{}},{"cell_type":"markdown","source":"## Kết quả full BoW đã lọc stop word, chưa có resampling\nTrain accuracy: 95.68%, F1: 0.5753\n\nTest accuracy:  95.09%, F1: 0.5107\n\n## Kết quả reduced BoW đã lọc stop word, chưa có resampling, có cài đặt giới hạn df (min = 0.0001, max = 0.999)\nTrain accuracy: 95.1808%, F1: 0.5061\n\nTest accuracy:  95.0675%, F1: 0.4945\n\n## Kết quả reduced BoW đã lọc stop word, chưa có resampling, chưa cài đặt giới hạn df\nTrain accuracy: 99.0091%, F1: 0.9140\n\nTest accuracy:  94.6525%, F1: 0.3530\n\n## Kết quả full BoW đã lọc stop word, đã có resampling với tỉ lệ 2:1\nTrain accuracy: 91.92%, F1: 0.8757\n\nTest accuracy:  87.85%, F1: 0.8099\n\n## Kết quả reduced BoW đã lọc stop word, đã có resampling với tỉ lệ 2:1\nTrain accuracy: 77.9344%, F1: 0.5636\n\nTest accuracy:  77.1047%, F1: 0.5435\n\n## Kết quả full BoW đã lọc stop word, đã có resampling với tỉ lệ 3:1\nTrain accuracy: 92.18%, F1: 0.8368\n\nTest accuracy:  89.29%, F1: 0.7720\n\n## Kết quả reduced BoW đã lọc stop word, đã có resampling với tỉ lệ 3:1\nTrain accuracy: 81.7226%, F1: 0.4973\n\nTest accuracy:  81.3003%, F1: 0.4753\n\n## Kết quả full BoW đã lọc stop word, đã có resampling với tỉ lệ 4:1\nTrain accuracy: 92.74%, F1: 0.8065\n\nTest accuracy:  90.35%, F1: 0.7409\n\n## Kết quả reduced BoW đã lọc stop word, đã có resampling với tỉ lệ 4:1\nTrain accuracy: 84.3101%, F1: 0.4372\n\nTest accuracy:  84.0069%, F1: 0.4279\n\n\n## Kết quả full BoW của Decision Tree, đã có resampling với tỉ lệ 4:1\nTraining accuracy: 88.28%, F1: 0.6437\n\nTesting accuracy:  86.95%, F1: 0.6097\n\n## Kết quả full BoW đã lọc stop word, đã có resampling với tỉ lệ 5:1\nTrain accuracy: 92.98%, F1: 0.7710\n\nTest accuracy:  91.06%, F1: 0.7053\n\n## Kết quả reduced BoW đã lọc stop word, đã có resampling với tỉ lệ 5:1\nTrain accuracy: 86.3180%, F1: 0.3913\n\nTest accuracy:  85.9021%, F1: 0.3726\n\n## Kết quả full BoW của Decision Tree, đã có resampling với tỉ lệ 5:1\nTraining accuracy: 89.75%, F1: 0.6242\n\nTesting accuracy:  88.35%, F1: 0.5766\n","metadata":{}},{"cell_type":"markdown","source":"# 6. Nhận xét kết quả đạt được\n\n* Có thể thấy rõ sự khác biệt giữa các kết quả huấn luyện với các cách xử lý dữ liệu khác nhau, trong đó, rõ nhất là khi dữ liệu được resampling và không được resampling: Khi resampling dữ liệu, ta thu được điểm F1 tăng tới hơn 60% so với khi không resampling dữ liệu.\n* Tuy nhiên, quá trình lọc stop word không thực sự hữu ích, và điểm số so với những lần không lọc stop word còn thấp hơn.\n* Bộ dữ liệu Full BoW cho kết quả ấn tượng hơn nhiều so với Reduced BoW, cả về độ chính xác lẫn điểm F1. Ngoài ra, khoảng cách giữa độ chính xác và điểm F1 của Reduced BoW quá xa (có thể chênh nhau tới 0.2-0.4 điểm), vậy nên Reduced BoW sẽ không được đưa vào nhận xét và sẽ nộp bài cho Kaggle bằng kết quả của Full BoW.\n* Khi lọc stop words và sử dụng resampling, thuật toán tiêu thụ ít bộ nhớ hơn và có thời gian huấn luyện nhanh hơn là không lọc stopwords (chỉ mất khoảng 6 phút để huấn luyện so với 15 phút trước đây).\n* Khi tăng tỉ lệ resampling giữa số lượng câu hỏi chân thành : không chân thành, điểm số F1 và độ chính xác của mô hình có xu hướng giảm, nhưng điểm số huấn luyện của Kaggle có xu hướng tăng (Cụ thể trong cách tiếp cận này, tỉ lệ 5:1 là tỉ lệ resampling cho ra kết quả Kaggle cao nhất).\n","metadata":{}},{"cell_type":"markdown","source":"# 7. Đánh giá hiệu năng và nộp bài cho Kaggle\n\n* Resampling 1:1: 0.47890 private, 0.47077 public\n* Resampling 2:1: 0.55054 private, 0.54269 public\n* Resampling 3:1: 0.57915 private, 0.56736 public\n* Resampling 4:1: 0.58155 private, 0.57650 public\n* Resampling 5:1: 0.58816 private, 0.58500 public\n\n#### Nhận xét chung: Tuy kết quả huấn luyện đưa ra điểm số khá tự tin, khi ta đưa một bộ dữ liệu mới để kiểm tra thì mô hình đưa ra kết quả không khả thấp hơn dự kiến. Vậy nên mô hình huấn luyện này đã mắc phải lỗi \"Overfitting\" - là khi mô hình đưa ra kết quả chính xác với một tập dữ liệu cụ thể, nhưng không thể đưa ra kết quả chính xác khi bổ sung thêm dữ liệu mới.\n","metadata":{}},{"cell_type":"markdown","source":"#### Đoạn code bên dưới phục vụ mục đích đưa ra kết quả của mô hình đối với bộ dữ liệu ```test```, và sẽ dùng kết quả đó để nộp bài cho Kaggle.","metadata":{}},{"cell_type":"code","source":"final_vectorizer = CountVectorizer()\nfinal_model = LogisticRegression(C=3.5)\n\nfinal_bow = final_vectorizer.fit_transform(all_texts) \nfinal_model.fit(final_bow, target_array.values) \nfinal_train_predictions = final_model.predict(final_bow) \nfinal_acc = accuracy_score(target_array.values, final_train_predictions) \nfinal_f1 = f1_score(target_array.values, final_train_predictions) \nprint(f\"Final model accuracy:  {final_acc:.2%}, F1: {final_f1:.4f}\") \n","metadata":{"execution":{"iopub.status.busy":"2021-06-11T01:41:07.466076Z","iopub.execute_input":"2021-06-11T01:41:07.466571Z","iopub.status.idle":"2021-06-11T01:41:28.258888Z","shell.execute_reply.started":"2021-06-11T01:41:07.466527Z","shell.execute_reply":"2021-06-11T01:41:28.257920Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#### Xuất ra tệp ```submission.csv``` để nộp lên Kaggle. Kết quả chấm điểm từ bên Kaggle đã được ghi rõ ở mục phía trên.","metadata":{"execution":{"iopub.status.busy":"2021-06-09T10:51:17.433913Z","iopub.execute_input":"2021-06-09T10:51:17.434334Z","iopub.status.idle":"2021-06-09T10:54:19.071358Z","shell.execute_reply.started":"2021-06-09T10:51:17.434287Z","shell.execute_reply":"2021-06-09T10:54:19.070142Z"}}},{"cell_type":"code","source":"validation_texts = np.array([process(x) for x in test_data['question_text']])\nvalidation_bow = final_vectorizer.transform(validation_texts) \nvalidation_predictions = final_model.predict(validation_bow) \nsubmission = pd.DataFrame({'qid':test_data['qid'], 'prediction':validation_predictions })\nsubmission.to_csv('./submission.csv', index=False)","metadata":{"execution":{"iopub.status.busy":"2021-06-11T01:41:28.260563Z","iopub.execute_input":"2021-06-11T01:41:28.261102Z","iopub.status.idle":"2021-06-11T01:43:29.817927Z","shell.execute_reply.started":"2021-06-11T01:41:28.261060Z","shell.execute_reply":"2021-06-11T01:43:29.816933Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# 8. Tổng kết\nBài báo cáo này đã đưa ra hướng tiếp cận để giải quyết bài toán lọc câu hỏi không chân thành của Quora. Bài báo cáo đã nêu rõ vấn đề, phân tích dữ liệu, độ nhiễu để đề ra phương hướng xử lý, cũng như là cân bằng lại dữ liệu. Dữ liệu sau khi xử lý được phân tách thành các ```Bag-of-word```, trước khi cho vào thuật toán học máy để huấn luyện mô hình. Kết quả đầu ra được kiểm chứng một lần nữa bởi bộ dữ liệu kiểm thử cuối, trước khi được xuất ra tệp ```csv``` để nộp chấm điểm trên Kaggle.","metadata":{}}]}