Skip to content

A sample neural network created without any Deep Learning library (pytorch, tensorflow etc) to show the basics of how deep learning works.

License

Notifications You must be signed in to change notification settings

DevUnleash/Predicting-Bike-Sharing

Repository files navigation

Predicting Bike Sharing

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.

Installing dependencies

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

Unit Tests

To run the unit tests, you can simply execute pyton network-test.py

Running the project

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.

Contributing

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.

License

MIT

About

A sample neural network created without any Deep Learning library (pytorch, tensorflow etc) to show the basics of how deep learning works.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published