Predictive-Analysis

The repository contains the code for the various machine learning algorithms used to make a predictive analysis of tweets on GST in India The Goods and Services Tax has been a revolutionary change in the financial standards of India. This led to a widespread debate across all social media platforms on the severity of its effects on the common man of this country. The very fact that many reactions collected on the social media regarding this topic has brought about the need to bifurcate the reactions based on their sentiments. Our goal is to not only classify the reactions based on the sentiments, but also to predict whether the upcoming tweets on this issue is on a positive note, or a negative note. This analysis can be done by classifying the dataset using various Machine Learning Algorithms. The main goal is to conclude, which of the models used is the most accurate in predicting the outcome. Classification models used : Linear Regression, Support Vector Machines, Naïve Bayes, Decision Tree, Random Forest, XGBoost, RNN-LSTM. Receiver Operating characteristic curves, Word Embeddings. The file tweet.py contains the code to extract tweets from twitter related to GST from twitter API. From the nltk library we stem the stop words such as 'the', 'if', 'them' etc. The files gst_decesiontree.py, gst_kn.py, gst_logi.py, etc are the various machine learning algorithms that can be used for the predictive analysis. Confusion matrices containing values of true positive, true negative, false postives and false negatives give the accuracy of the predicted sentiments. To visualise the accuracy of the results ROC - Receiver Operating Characterstic Curves were plotted using the matplotlib.pyplot were plotted. The files ending with roc.py are the roc plots for the respective models.