pip install vaderSentiment

import pandas as pd import numpy as np from textblob import TextBlob import re import nltk from vaderSentiment.vaderSentiment import SentimentIntensityAnalyzer from sklearn.model_selection import train_test_split from sklearn.metrics import accuracy_score, classification_report from sklearn.discriminant_analysis import LinearDiscriminantAnalysis

from google.colab import files files.upload()

df1 = pd.read_csv('datasets_129_792900_upload_DJIA_table.csv') df2 = pd.read_csv('129_792900_compressed_Combined_News_DJIA.csv.zip')

df1.head(3)

df2.head(3)

merge = df1.merge(df2, how='inner', on='Date', left_index = True) merge.head(3)

headlines = [] for row in range(0,len(merge.index)): headlines.append(' '.join(str(x) for x in merge.iloc[row,2:27]))

clean_headlines = [] for i in range(0, len(headlines)): clean_headlines.append(re.sub("b[(')]+", '', headlines[i] )) clean_headlines[i] = re.sub('b[(")]+', '', clean_headlines[i] ) clean_headlines[i] = re.sub("'", '', clean_headlines[i] )

merge['Combined_News'] = clean_headlines

def getSubjectivity(text): return TextBlob(text).sentiment.subjectivity

def getPolarity(text): return TextBlob(text).sentiment.polarity

merge['Subjectivity'] =merge['Combined_News'].apply(getSubjectivity) merge['Polarity'] =merge['Combined_News'].apply(getPolarity)

def getSIA(text): sia = SentimentIntensityAnalyzer() sentiment = sia.polarity_scores(text) return sentiment

compound = [] neg = [] neu = [] pos = [] SIA = 0 for i in range(0, len(merge['Combined_News'])): SIA = getSIA(merge['Combined_News'][i]) compound.append(SIA['compound']) neg.append(SIA['neg']) neu.append(SIA['neu']) pos.append(SIA['pos'])

merge['Compound'] =compound merge['Negative'] =neg merge['Neutral'] =neu merge['Positive'] = pos

keep_columns = [ 'Open', 'High', 'Low', 'Volume', 'Subjectivity', 'Polarity', 'Compound', 'Negative', 'Neutral' ,'Positive', 'Label' ] df = merge[keep_columns] df

X = df X = np.array(X.drop(['Label'], 1))

y = np.array(df['Label'])

x_train, x_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state = 0)

model = LinearDiscriminantAnalysis().fit(x_train, y_train)

predictions = model.predict(x_test) predictions

print( classification_report(y_test, predictions) )