3 layer neural network class with gradient descent backpropogation written only in numpy. Environmental and seasonal features used to predict bike share rental ridership in Washington, D.C.

Aggregated bike sharing rental and corresponding weather/seasonal information from http://capitalbikeshare.com/system-data and http://www.freemeteo.com. Compiled by University of Porto, Portugal and distributed by Udacity.

The NeuralNetwork class in nn.py has 5 user defined variables. 1. input_odes: the number of features that you wish to train on 2. hidden_nodes: the number of hidden nodes in the hidden layer 3. output_nodes: the number of output nodes 4. learning_rate: scaling for backpropogated weights (adaptability to new data) 5. iterations: number of iterations

To try it, make sure requirements.txt are met. Then, run eda.py and Bike-Sharing-Dataset. Adjust hyperparameters

fix the plots and add them above