carsoli / prediction-tasks Goto Github PK

This project is divided intro three tasks as per the included description PDF. In the first task, a NN model was used to predict the value of the S&P Index based on company stock prices. Initially, the data-set included stock prices of 387 companies, but only the ones with high correlation with the S&P Index were used. Grid Search was performed to obtain the best configuration for hyper-parameters. In the second task, time series data of the stock prices was constructed using the data-set from the first task, and the prediction was based on a 4-day window, and predicted 4 days ahead. The third task was a NLP task, where ratings for products from an Amazon data-set are predicted based on textual reviews. Starter code was provided for the pre-processing of the text, and is included as well.