This project is a solution of Udacity's Deep learning Nanodegree project. In this project, we've created a sample neural network and used it to predict daily bike rental ridership. We didn't use any DL framework (tensorflow, pytorch etc) just to demonstrate the ideas behind the basic concepts in neural networks. The neural network implementation in this kernel elaborates the concepts like forward and backward propagation, stochastic gradient descent and hyperparameters' explanations.
This is a python 3 project. You need to install python3 first and then you can install the dependencies using:
pip install -r requirements.txt
To run the unit tests, you can simply execute pyton network-test.py
run.py is the main executable script that loads the data, trains a neural network and runs a test to predict bike sharing. You can simply run python run.py to train the network and test it.
You can contribute to this project by opening a pull request. Specifically, you can try out different neural network implementations. Also, you can fork this kernel on kaggle.
MIT