{"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":"markdown","source":"

Natural Language Processing, EDA & Prediction of Tweets

\n","metadata":{}},{"cell_type":"code","source":"import numpy as np \nimport pandas as pd \nimport nltk\nfrom nltk.tokenize import word_tokenize, TweetTokenizer\nfrom nltk.corpus import stopwords\nfrom nltk.stem import WordNetLemmatizer, PorterStemmer\nfrom nltk.probability import FreqDist\nimport string as s\nimport re\nfrom textblob import TextBlob\nimport matplotlib.pyplot as plt\nfrom sklearn.decomposition import TruncatedSVD\nfrom sklearn.model_selection import cross_validate, GridSearchCV, train_test_split\nfrom sklearn.naive_bayes import MultinomialNB\nfrom sklearn.linear_model import LogisticRegression\nfrom sklearn.neighbors import KNeighborsClassifier\nfrom catboost import CatBoostClassifier\nfrom xgboost import XGBClassifier\nfrom sklearn.ensemble import RandomForestClassifier, GradientBoostingClassifier, AdaBoostClassifier\nfrom sklearn.metrics import roc_auc_score, roc_curve, confusion_matrix\nfrom sklearn.compose import ColumnTransformer\nfrom sklearn.pipeline import Pipeline, make_pipeline\nfrom sklearn.feature_extraction.text import TfidfVectorizer, CountVectorizer\nfrom sklearn.preprocessing import StandardScaler, MinMaxScaler\nfrom sklearn.utils.class_weight import compute_class_weight\nfrom skopt import BayesSearchCV\nfrom skopt.space import Real, Categorical, Integer\nimport os\nfor dirname, _, filenames in os.walk('/kaggle/input'):\n for filename in filenames:\n print(os.path.join(dirname, filename))","metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","execution":{"iopub.status.busy":"2022-07-27T23:51:09.123195Z","iopub.execute_input":"2022-07-27T23:51:09.123514Z","iopub.status.idle":"2022-07-27T23:52:26.724100Z","shell.execute_reply.started":"2022-07-27T23:51:09.123460Z","shell.execute_reply":"2022-07-27T23:52:26.722989Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"

Load Dataset & Get an informations about it

","metadata":{}},{"cell_type":"code","source":"train_df=pd.read_csv('/kaggle/input/nlp-getting-started/train.csv')\ntest_df=pd.read_csv('/kaggle/input/nlp-getting-started/test.csv')\ntrain_mod_df = train_df.copy()\ntest_mod_df = test_df.copy()\ntest_df2 = test_df.copy()\ntrain_df2 = train_df.copy()\ndef data_info(d):\n print('number of variables: ',d.shape[1])\n print('number of tweets: ',d.shape[0])\n print('variables names: ')\n print(d.columns)\n print('variables data-types: ')\n print(d.dtypes)\n print('missing values: ')\n c=d.isnull().sum()\n print(c[c>0])","metadata":{"_uuid":"d629ff2d2480ee46fbb7e2d37f6b5fab8052498a","_cell_guid":"79c7e3d0-c299-4dcb-8224-4455121ee9b0","execution":{"iopub.status.busy":"2022-07-20T22:39:13.005715Z","iopub.execute_input":"2022-07-20T22:39:13.006060Z","iopub.status.idle":"2022-07-20T22:39:13.047338Z","shell.execute_reply.started":"2022-07-20T22:39:13.006007Z","shell.execute_reply":"2022-07-20T22:39:13.046588Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"data_info(train_df)","metadata":{"execution":{"iopub.status.busy":"2022-07-20T22:39:16.441420Z","iopub.execute_input":"2022-07-20T22:39:16.441716Z","iopub.status.idle":"2022-07-20T22:39:16.456719Z","shell.execute_reply.started":"2022-07-20T22:39:16.441666Z","shell.execute_reply":"2022-07-20T22:39:16.456121Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"data_info(test_df)","metadata":{"execution":{"iopub.status.busy":"2022-07-20T22:39:19.575788Z","iopub.execute_input":"2022-07-20T22:39:19.576104Z","iopub.status.idle":"2022-07-20T22:39:19.587352Z","shell.execute_reply.started":"2022-07-20T22:39:19.576051Z","shell.execute_reply":"2022-07-20T22:39:19.585948Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# remove location from test and training datasets\ntrain_mod_df.drop(['location', 'id', 'keyword'], axis = 1, inplace = True)\ntest_mod_df.drop(['location', 'id', 'keyword'], axis = 1, inplace = True)\ntrain_df2.drop(['location', 'id', 'keyword'], axis = 1, inplace = True)\ntest_df2.drop(['location', 'id', 'keyword'], axis = 1, inplace = True)","metadata":{"execution":{"iopub.status.busy":"2022-07-20T22:39:21.264169Z","iopub.execute_input":"2022-07-20T22:39:21.264468Z","iopub.status.idle":"2022-07-20T22:39:21.276339Z","shell.execute_reply.started":"2022-07-20T22:39:21.264415Z","shell.execute_reply":"2022-07-20T22:39:21.273651Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"

EDA

","metadata":{}},{"cell_type":"code","source":"plt.figure(figsize=(8,8))\ntrain_df['target'].value_counts().plot.pie(autopct='%.2f%%')\nplt.title('Disaster or Not Distribution')\nplt.ylabel('')\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2022-07-20T22:39:26.826864Z","iopub.execute_input":"2022-07-20T22:39:26.827148Z","iopub.status.idle":"2022-07-20T22:39:26.957588Z","shell.execute_reply.started":"2022-07-20T22:39:26.827102Z","shell.execute_reply":"2022-07-20T22:39:26.956794Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_mod_df[train_mod_df['target']==1].loc[:4,'text']","metadata":{"execution":{"iopub.status.busy":"2022-07-20T22:39:28.835839Z","iopub.execute_input":"2022-07-20T22:39:28.836127Z","iopub.status.idle":"2022-07-20T22:39:28.847616Z","shell.execute_reply.started":"2022-07-20T22:39:28.836077Z","shell.execute_reply":"2022-07-20T22:39:28.846634Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_disasters={'earthquake':0,'fire':0,'fires':0,'shelter':0}\ntest_disasters={'earthquake':0,'fire':0,'fires':0,'shelter':0}\nfor i,j in zip(train_df['text'],test_df['text']):\n if 'earthquake' in i.split():\n train_disasters['earthquake']+=1\n elif 'earthquake' in j.split():\n test_disasters['earthquake']+=1\n if 'fire' in i.split():\n train_disasters['fire']+=1\n elif 'fire' in j.split():\n test_disasters['fire']+=1\n if 'fires' in i.split():\n train_disasters['fires']+=1\n elif 'fires' in j.split():\n test_disasters['fires']+=1\n if 'shelter' in i.split():\n train_disasters['shelter']+=1\n elif 'shelter' in j.split():\n test_disasters['shelter']+=1\nprint('number of tweets that fires,fire and earthquacke mentioned in train data: ',train_disasters)\nprint('number of tweets that fires,fire and earthquacke mentioned in test data: ',test_disasters)","metadata":{"execution":{"iopub.status.busy":"2022-07-20T22:39:31.153297Z","iopub.execute_input":"2022-07-20T22:39:31.153585Z","iopub.status.idle":"2022-07-20T22:39:31.194399Z","shell.execute_reply.started":"2022-07-20T22:39:31.153536Z","shell.execute_reply":"2022-07-20T22:39:31.193131Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"

Text Preprocessing

","metadata":{}},{"cell_type":"code","source":"def remove_links_hash(s):\n s = re.sub(r'https?://[^\\s\\n\\r]+', '', s)\n s = re.sub(r'http?://[^\\s\\n\\r]+', '', s)\n s = re.sub(r'\\#\\w+', '', s)\n s = re.sub(r'\\@\\w+', '', s) \n return s\ndef remove_emojis(s):\n emoj = re.compile(\"[\"\n u\"\\U0001F600-\\U0001F64F\" # emoticons\n u\"\\U0001F300-\\U0001F5FF\" # symbols & pictographs\n u\"\\U0001F680-\\U0001F6FF\" # transport & map symbols\n u\"\\U0001F1E0-\\U0001F1FF\" # flags (iOS)\n u\"\\U00002500-\\U00002BEF\" # chinese char\n u\"\\U00002702-\\U000027B0\"\n u\"\\U00002702-\\U000027B0\"\n u\"\\U000024C2-\\U0001F251\"\n u\"\\U0001f926-\\U0001f937\"\n u\"\\U00010000-\\U0010ffff\"\n u\"\\u2640-\\u2642\" \n u\"\\u2600-\\u2B55\"\n u\"\\u200d\"\n u\"\\u23cf\"\n u\"\\u23e9\"\n u\"\\u231a\"\n u\"\\ufe0f\" # dingbats\n u\"\\u3030\"\n \"]+\", re.UNICODE)\n return re.sub(emoj, '', s)\ntrain_mod_df['text'] = train_mod_df['text'].apply(remove_links_hash)\ntrain_mod_df['text'] = train_mod_df['text'].apply(remove_emojis)\ntest_mod_df['text'] = test_mod_df['text'].apply(remove_links_hash)\ntest_mod_df['text'] = test_mod_df['text'].apply(remove_emojis)","metadata":{"execution":{"iopub.status.busy":"2022-07-20T22:39:33.574155Z","iopub.execute_input":"2022-07-20T22:39:33.574458Z","iopub.status.idle":"2022-07-20T22:39:33.868877Z","shell.execute_reply.started":"2022-07-20T22:39:33.574406Z","shell.execute_reply":"2022-07-20T22:39:33.867859Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# string tokenizaer\ntt = TweetTokenizer()\ntrain_mod_df['text'] = train_mod_df['text'].apply(tt.tokenize)\ntest_mod_df['text'] = test_mod_df['text'].apply(tt.tokenize)\n# lowercase all words\ndef lowercasing(lst):\n new_lst=[]\n for i in lst:\n i = i.lower()\n new_lst.append(i)\n return new_lst\n# remove punctuations\ndef remove_punctuations(lst):\n new_lst=[]\n for i in lst:\n for j in s.punctuation:\n i=i.replace(j,'')\n new_lst.append(i)\n return new_lst\n# remove numbers\ndef remove_numbers(lst):\n nodig_lst=[]\n new_lst=[]\n for i in lst:\n for j in s.digits: \n i=i.replace(j,'')\n nodig_lst.append(i)\n for i in nodig_lst:\n if i!='':\n new_lst.append(i)\n return new_lst\ndef remove_stopwords(lst):\n stop=stopwords.words('english')\n new_lst=[]\n for i in lst:\n if i not in stop:\n new_lst.append(i)\n return new_lst\n# remove spaces\ndef remove_spaces(lst):\n new_lst=[]\n for i in lst:\n i=i.strip()\n new_lst.append(i)\n return new_lst\ndef remove_letters(lst):\n new_lst = []\n for i in lst:\n if len(i) > 2:\n new_lst.append(i)\n return new_lst\ndef remove_noisy_words(lst):\n noise_lst = ['aa',\n 'aa','ayyo',\n 'aa',\n 'aa', 'mgm',\n 'aa',\n 'aaaa',\n 'aaaa',\n 'aaaaaaallll',\n 'aaaaaaallll ûªm',\n 'aaaaaand',\n 'aaaaaand', 'theres',\n 'aaarrrgghhh',\n 'aal',\n 'aan',\n 'aan den',\n 'aannnnd',\n 'aannnnd',\n 'aar', 'ab']\n new_lst = []\n for word in lst:\n if word not in noise_lst:\n new_lst.append(word)\n return new_lst\n \ntrain_mod_df['text']=train_mod_df['text'].apply(lowercasing)\ntest_mod_df['text']=test_mod_df['text'].apply(lowercasing) \ntrain_mod_df['text']=train_mod_df['text'].apply(remove_punctuations)\ntest_mod_df['text']=test_mod_df['text'].apply(remove_punctuations) \ntrain_mod_df['text']=train_mod_df['text'].apply(remove_numbers)\ntest_mod_df['text']=test_mod_df['text'].apply(remove_numbers) \ntrain_mod_df['text']=train_mod_df['text'].apply(remove_stopwords)\ntest_mod_df['text']=test_mod_df['text'].apply(remove_stopwords) \ntrain_mod_df['text']=train_mod_df['text'].apply(remove_spaces)\ntest_mod_df['text']=test_mod_df['text'].apply(remove_spaces)\ntrain_mod_df['text']=train_mod_df['text'].apply(remove_letters)\ntest_mod_df['text']=test_mod_df['text'].apply(remove_letters)\ntrain_mod_df['text']=train_mod_df['text'].apply(remove_noisy_words)\ntest_mod_df['text']=test_mod_df['text'].apply(remove_noisy_words)","metadata":{"execution":{"iopub.status.busy":"2022-07-20T22:39:38.036013Z","iopub.execute_input":"2022-07-20T22:39:38.036307Z","iopub.status.idle":"2022-07-20T22:39:42.192390Z","shell.execute_reply.started":"2022-07-20T22:39:38.036254Z","shell.execute_reply":"2022-07-20T22:39:42.191568Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# lemmatization\nlemmatizer=WordNetLemmatizer()\ndef lemmatzation(lst):\n new_lst=[]\n for i in lst:\n i=lemmatizer.lemmatize(i)\n new_lst.append(i)\n return new_lst\ntrain_mod_df['text_lem']=train_mod_df['text'].apply(lemmatzation)\ntest_mod_df['text_lem']=test_mod_df['text'].apply(lemmatzation)","metadata":{"execution":{"iopub.status.busy":"2022-07-20T22:39:47.246487Z","iopub.execute_input":"2022-07-20T22:39:47.246775Z","iopub.status.idle":"2022-07-20T22:39:47.740563Z","shell.execute_reply.started":"2022-07-20T22:39:47.246723Z","shell.execute_reply":"2022-07-20T22:39:47.739559Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Stemming\nStemmer=PorterStemmer()\ndef stemmer(lst):\n new_lst=[]\n for i in lst:\n i=Stemmer.stem(i)\n new_lst.append(i)\n return new_lst\ntrain_mod_df['text_stem']=train_mod_df['text'].apply(stemmer)\ntest_mod_df['text_stem']=test_mod_df['text'].apply(stemmer)","metadata":{"execution":{"iopub.status.busy":"2022-07-20T22:39:51.272341Z","iopub.execute_input":"2022-07-20T22:39:51.272686Z","iopub.status.idle":"2022-07-20T22:39:53.662127Z","shell.execute_reply.started":"2022-07-20T22:39:51.272633Z","shell.execute_reply":"2022-07-20T22:39:53.661268Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"

Converting Text to Features(Vectorization) and Modelling

","metadata":{}},{"cell_type":"code","source":"train_mod_df['text']=train_mod_df['text'].apply(lambda x: ''.join(i+' ' for i in x)).str.rstrip()\ntest_mod_df['text']=test_mod_df['text'].apply(lambda x: ''.join(i+' ' for i in x)).str.rstrip()\ntrain_mod_df['text_lem']=train_mod_df['text_lem'].apply(lambda x: ''.join(i+' ' for i in x)).str.rstrip()\ntrain_mod_df['text_stem']=train_mod_df['text_stem'].apply(lambda x: ''.join(i+' ' for i in x)).str.rstrip()\ntest_mod_df['text_lem']=test_mod_df['text_lem'].apply(lambda x: ''.join(i+' ' for i in x)).str.rstrip()\ntest_mod_df['text_stem']=test_mod_df['text_stem'].apply(lambda x: ''.join(i+' ' for i in x)).str.rstrip()\ntrain_mod_df.head()","metadata":{"execution":{"iopub.status.busy":"2022-07-20T22:41:23.413306Z","iopub.execute_input":"2022-07-20T22:41:23.413629Z","iopub.status.idle":"2022-07-20T22:41:23.528264Z","shell.execute_reply.started":"2022-07-20T22:41:23.413573Z","shell.execute_reply":"2022-07-20T22:41:23.526961Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"count_vec = CountVectorizer(ngram_range = (1, 1))\ntfidf_vec = TfidfVectorizer(ngram_range = (1, 1))\ncount_vec.fit(train_mod_df['text_stem'])\nX_stem_count = count_vec.transform(train_mod_df['text_stem'])\ntest_stem_count = count_vec.transform(test_mod_df['text_stem'])\ntfidf_vec.fit(train_mod_df['text_stem'])\nX_stem_tfidf = tfidf_vec.transform(train_mod_df['text_stem'])\ntest_stem_tfidf = tfidf_vec.transform(test_mod_df['text_stem'])\n'''\n#####\ncount_vec2 = CountVectorizer(ngram_range = (1, 2))\ntfidf_vec2 = TfidfVectorizer(ngram_range = (1, 2))\n####\ncount_vec3 = CountVectorizer(ngram_range = (1, 2))\ntfidf_vec3 = TfidfVectorizer(ngram_range = (1, 2))\nX_count = count_vec.fit_transform(X)\ntest_mod_df_count = count_vec.transform(test_mod_df['text'])\n###\nX_lem_count = count_vec2.fit_transfrom(X_lem)\ntest_lem_count = count_vec2.transform(test_lem)\n###\nX_stem_count = count_vec3.fit_transfrom(X_stem)\ntest_stem_count = count_vec3.transform(test_stem)\n# TF-IDF\nX_tfidf = tfidf_vec.fit_transform(X)\ntest_mod_df_tfidf = tfidf_vec.transform(test_mod_df['text'])\n##\nX_lem_tfidf = tfidf_vec2.fit_transfrom(X_lem)\ntest_lem_tfidf = tfidf_vec2.transform(test_lem)\n##\nX_stem_tfidf = tfidf_vec3.fit_transfrom(X_stem)\ntest_stem_tfidf = tfidf_vec3.transform(test_stem)\n'''","metadata":{"execution":{"iopub.status.busy":"2022-07-21T00:02:04.029200Z","iopub.execute_input":"2022-07-21T00:02:04.029495Z","iopub.status.idle":"2022-07-21T00:02:04.588364Z","shell.execute_reply.started":"2022-07-21T00:02:04.029446Z","shell.execute_reply":"2022-07-21T00:02:04.587727Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"X_tfidf_train = pd.DataFrame(np.log1p(X_stem_tfidf.todense()), \n columns=tfidf_vec.get_feature_names())\nX_tfidf_test = pd.DataFrame(np.log1p(test_stem_tfidf.todense()), \n columns=tfidf_vec.get_feature_names())","metadata":{"execution":{"iopub.status.busy":"2022-07-21T00:02:09.431312Z","iopub.execute_input":"2022-07-21T00:02:09.431623Z","iopub.status.idle":"2022-07-21T00:02:10.508217Z","shell.execute_reply.started":"2022-07-21T00:02:09.431566Z","shell.execute_reply":"2022-07-21T00:02:10.507351Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"svd = TruncatedSVD(n_components = 200)\nsvd_train = svd.fit_transform(X_tfidf_train)\nsvd_test = svd.transform(X_tfidf_test)","metadata":{"execution":{"iopub.status.busy":"2022-07-21T00:02:19.691913Z","iopub.execute_input":"2022-07-21T00:02:19.692205Z","iopub.status.idle":"2022-07-21T00:02:32.468343Z","shell.execute_reply.started":"2022-07-21T00:02:19.692156Z","shell.execute_reply":"2022-07-21T00:02:32.467324Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"y = train_mod_df['target']","metadata":{"execution":{"iopub.status.busy":"2022-07-20T23:34:25.524525Z","iopub.execute_input":"2022-07-20T23:34:25.524816Z","iopub.status.idle":"2022-07-20T23:34:25.529200Z","shell.execute_reply.started":"2022-07-20T23:34:25.524755Z","shell.execute_reply":"2022-07-20T23:34:25.528119Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# now let's calculate accuracy, f1_score, precision, recall and auc from cross validation.\n# also calculate the training time.\nimport time\ndef calc_cv_scores(model, X, y, cv = 5):\n start = time.time()\n scores = cross_validate(model, X,\n y,\n cv = cv,\n n_jobs = -1,\n scoring = ['accuracy',\n 'f1',\n 'precision',\n 'recall',\n 'roc_auc'])\n end = time.time()\n print(f'mean accuracy: {np.mean(scores[\"test_accuracy\"])}')\n print(f'mean f1: {np.mean(scores[\"test_f1\"])}')\n print(f'mean precision: {np.mean(scores[\"test_precision\"])}')\n print(f'mean recall: {np.mean(scores[\"test_recall\"])}')\n print(f'mean AUC: {np.mean(scores[\"test_roc_auc\"])}')\n print(f'training time: {end - start}')","metadata":{"execution":{"iopub.status.busy":"2022-07-20T23:30:50.327715Z","iopub.execute_input":"2022-07-20T23:30:50.328049Z","iopub.status.idle":"2022-07-20T23:30:50.335763Z","shell.execute_reply.started":"2022-07-20T23:30:50.327998Z","shell.execute_reply":"2022-07-20T23:30:50.334939Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"calc_cv_scores(LogisticRegression(), svd_train, y, cv = 5)","metadata":{"execution":{"iopub.status.busy":"2022-07-21T00:02:58.923548Z","iopub.execute_input":"2022-07-21T00:02:58.923869Z","iopub.status.idle":"2022-07-21T00:03:01.553368Z","shell.execute_reply.started":"2022-07-21T00:02:58.923794Z","shell.execute_reply":"2022-07-21T00:03:01.552603Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"class_weights = compute_class_weight(class_weight = 'balanced', classes = np.unique(y), y = y)\nclass_weights","metadata":{"execution":{"iopub.status.busy":"2022-07-20T23:48:45.739716Z","iopub.execute_input":"2022-07-20T23:48:45.740074Z","iopub.status.idle":"2022-07-20T23:48:45.749258Z","shell.execute_reply.started":"2022-07-20T23:48:45.740021Z","shell.execute_reply":"2022-07-20T23:48:45.748317Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"calc_cv_scores(LogisticRegression(class_weight = {0:class_weights[0], 1:class_weights[1]}), svd_train, y, cv = 5)","metadata":{"execution":{"iopub.status.busy":"2022-07-21T00:03:09.567190Z","iopub.execute_input":"2022-07-21T00:03:09.567501Z","iopub.status.idle":"2022-07-21T00:03:10.036732Z","shell.execute_reply.started":"2022-07-21T00:03:09.567447Z","shell.execute_reply":"2022-07-21T00:03:10.035932Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"calc_cv_scores(RandomForestClassifier(), svd_train, y, cv = 5)","metadata":{"execution":{"iopub.status.busy":"2022-07-21T00:03:16.499165Z","iopub.execute_input":"2022-07-21T00:03:16.499479Z","iopub.status.idle":"2022-07-21T00:03:19.541448Z","shell.execute_reply.started":"2022-07-21T00:03:16.499426Z","shell.execute_reply":"2022-07-21T00:03:19.540646Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"'''\nmodel = [\n {\n 'name': 'AdaBoost',\n 'estimator': AdaBoostClassifier(random_state = 42),\n 'hyperparameters':{\n 'n_estimators' : [40, 80, 150, 400, 700, 1000],\n 'learning_rate':[0.01, 0.1, 0.3, 0.5, 0.8]\n }\n },\n {\n 'name':'Naive Bayes',\n 'estimator': MultinomialNB(),\n 'hyperparameters':{\n 'alpha': [0.1, 0.3, 0.5, 0.7, 0.9]\n }\n }\n]\nfor i in model:\n print(i['name'])\n bs=GridSearchCV(i['estimator'],\n param_grid=i['hyperparameters'],\n cv=5, n_jobs=-1,\n scoring='f1')\n bs.fit(X_features, y)\n print('best score: ', bs.best_score_)\n print('best parameters ; ', bs.best_params_)\n print('best model: ', bs.best_estimator_)\n'''","metadata":{"execution":{"iopub.status.busy":"2022-07-08T12:45:17.948646Z","iopub.execute_input":"2022-07-08T12:45:17.948949Z","iopub.status.idle":"2022-07-08T12:48:13.500142Z","shell.execute_reply.started":"2022-07-08T12:45:17.948902Z","shell.execute_reply":"2022-07-08T12:48:13.495963Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"

Prediction

","metadata":{}},{"cell_type":"code","source":"final_model = LogisticRegression(class_weight = {0:class_weights[0], 1:class_weights[1]})\n#final_model = RandomForestClassifier()\nfinal_model.fit(svd_train, y)\ny_pred = final_model.predict(svd_test)\ny_pred[:15]","metadata":{"execution":{"iopub.status.busy":"2022-07-21T00:03:32.940311Z","iopub.execute_input":"2022-07-21T00:03:32.940603Z","iopub.status.idle":"2022-07-21T00:03:33.060394Z","shell.execute_reply.started":"2022-07-21T00:03:32.940552Z","shell.execute_reply":"2022-07-21T00:03:33.059538Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"

Submission

","metadata":{}},{"cell_type":"code","source":"submission_df = {\"id\":test_df['id'],\n \"target\":y_pred}\nsubmission = pd.DataFrame(submission_df)\nsubmission.to_csv('submission_df.csv',index=False)","metadata":{"execution":{"iopub.status.busy":"2022-07-21T00:03:38.055485Z","iopub.execute_input":"2022-07-21T00:03:38.055781Z","iopub.status.idle":"2022-07-21T00:03:38.077853Z","shell.execute_reply.started":"2022-07-21T00:03:38.055723Z","shell.execute_reply":"2022-07-21T00:03:38.076901Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]}]}