Fake-News-Classification

• Dataset is taken from a US news agency containing 20k records with ’title’ and ’text’ as main features.

• An interesting combination of techniques for data preprocessing(BOW, TF-IDF, NLTK) and training(Naive Bayes, Random Forest, LSTM-RNN) are used.

• Although TF-IDF with Random Forest excelled all and gave the best accuracy of 94.31% but LSTM-RNN was time-efficient with an accuracy of 90.3%.

Chatbot-with-Python-and-Deep-Learning

• The input was given in the form of a manually created json file including “tag”, “pattern” and ‘‘responses” as main features. The file mainly included simple human Intents like ‘greetings’, ‘goodbye’, ‘about’, ‘help’, etc.

• Used Tokenizer from Tensorflow for text preprocessing and Word2Vec Embedding layer for Vectorization.

• The model was trained and saved(as a pickle file) to be used as a chatbot while interacting with the user.

Apple Stock Price Prediction using Stacked LSTM

• Predicted Apple stock prices for the next 30 days using LSTM based on last five years’ data (Time series data).

• Performed EDA and trained LSTM model using Keras.

• Predicted stock price successfully with minimal MSE loss.

Heart-Disease-Prediction-using-Machine-Learning

• Used a variety of ML algorithms, implemented in Python, to predict the presence of heart disease in a patient.

• Performed EDA, Data Preprocessing, and Model fitting using Logistic Regression, Naive Bayes, SVM, KNN, Decision Tree, Random Forest, XGboost.

• Random Forest performed best with an accuracy of 91.8%.

Other Projects

1. Stock-Sentiment-Analysis-using-Top-News-Headlines (NLP, Machine Learning, Naive Bayes)

2. Movie-Reviews-Sentiment-Analysis (NLP, Machine Learning, Naive Bayes)

3. Google-Stock-Price-Prediction-using-Timeseries-Data (Time series data, Deep Learning, LSTM RNN)

4. Prediction-of-bank-customer-s-satisfaction-using-ANN (Bank data, Deep Learning, ANN)

5. Spam-Message-Classification (NLP, Machine Learning, Naive Bayes)