This repo contains the code, data and output for the class project in Stats 205. The goal is to predict the number of bike users per day.

Contains code to reproduce outputs

• load_clean: R code to load, clean and pre-process the data (is sourced in methods.R)
• descriptives: R code to produce the descriptives (is sourced in methods.R)
• methods.R: R code to create forecasts based on all methods except LSTM
• lstm_forecasting.ipynb: Python code for forecasting using a LSTM NN

• subfolder plots: Contains all plots created. This includes descriptives as well as plots of the fitted and forecasted values for each method.
• subfolder tables: Contains summary statistics as well as a table comparing the RMSE for each method.

Contains extra packages.

Contains some papers and books related to this project.