Elon’s Tremendous Tweets

Natural Language Processor (NLP) selected from best performing model between logistic regression and LSVM models trained and tested on general tweets. The NLP was then used to perform sentiment analysis on Elon Musk’s tweets

Stock Forecasting

Time series forecasting of current MSFT stock closing price using autocorrelation to select number of time lags, and periodicity analysis to remove annual variation

Autoregressive integrated moving average (ARIMA) modeling used to project stock price close for MSFT over a 20 day window.

Sing Bot Sing

What is music? I certainly don’t know. Maybe a neural network will, or at least be able to predict musical notes.

This project consisted of decomposing MIDI data to a sequence of notes, durations, and offsets. Then, implementing multiple LSTM (long short term memory) network models to predict current note, duration of the note, and offset (where in the song the note is played) using a combination of time lagged features.

Finally, using the notes, offsets, and durations predicted by the model, building a MIDI file to be played in any music editor.