class sentiment:
    def sentiment():
       data = pd.read_csv("dump.csv")
       duplicate_data = data.copy()

        duplicate_data['reviews'] = duplicate_data['reviews__body']
        duplicate_data['sentiment'] = duplicate_data['reviews__label']

        duplicate_data = duplicate_data.loc[:, ['reviews__body','reviews__label']]

        train, test = train_test_split(duplicate_data, test_size=0.2, random_state = 1)
        X_train = train['reviews__body'].values
        X_test = test['reviews__body'].values
        y_train = train['reviews__label']
        y_test = test['reviews__label']

        def tokenize(text):
            return word_tokenize(text)

        def stem(doc):
            return (stemmer.stem(w) for w in analyzer(doc))

        get_stopwords = set(stopwords.words("english"))

        vectorizer = CountVectorizer(analyzer = 'word',tokenizer = tokenize,lowercase = True,ngram_range = (1,1),stop_words = get_stopwords)

        kfolds = StratifiedKFold(n_splits=5, shuffle=True, random_state=1)

        np.random.seed(1)

        pipeline_svm = make_pipeline(vectorizer, SVC(probability=True, kernel="linear", class_weight="balanced"))

        grid_svm = GridSearchCV(pipeline_svm, param_grid = {'svc_C': [0.01, 0.1, 1]}, cv = kfolds, scoring="roc_auc", verbose=1, n_jobs=-1)

        print(grid_svm.fit(X_train, y_train))