geiszla/leave-times

A simple machine learning project of predicting bus leave times using TensorFlow.

Leave Times

Overview

This project contains a basic python application for predicting bus leave times with different machine learning models (linear regression, DNN regression).

The project is highly modular, a different model can be used with the same data and training process by inheriting from the Model class and creating a new instance of it in leave_times/__init__.py (see included models for examples).

Requirements

• Python 3 (developed using v3.5.4)

Setting up

Make sure python, pip and the Scripts directory are in the path and run the following commands:

1. pip install --upgrade --user virtualenv
2. cd {project directory}
3. virtualenv env
4. env\Scripts\activate
5. pip install pip-tools

For development

6. pip-sync requirements-dev.txt

For production

6. pip-sync requirements.txt

Start application

Activate virtual environment (env\Scripts\activate), then run: python leave_times/__init__.py.

Project structure

• Models
  • Linear regression: leave_times/linear_regression.py
  • Neural Network: leave_times/neural_network.py
• Data
  • Training data: data/log.csv
  • Test data: data/log_unlabeled.csv

Credits

This project is based on Google's Machine Learning Crash Course.