This is an open source project aims at making a prediction of copper price using machine learning / deep learning approach. The language of this project is Python and the ideas may extend to other time series prediction problems as you like.
The motivation of launching this project is that copper is a kind of important raw material of some midstream and downstream materials, whose price is of great relevance to copper price. Thus a high accuracy of copper price prediction can to some extent providing purchasing decisions for the purchasing agent.
Here is a comparison of the fluctuation of the price of cooper and a kind of downstream material:
- Python 3.5 +
- numpy 1.13.0 +
- pandas 0.20.2 +
- scikit-learn 0.18.1 +
- Keras 2.0.4 +
- tensorflow 1.1.0 +
- matplotlib 2.0.2 +
- mysql-connector-python-rf 2.2.2 +
- common: common method such as data loading, model evaluation & model visualization etc.
- mlp: predict copper price using sklearn.neural_network.MLPRegressor
- lstm: predict copper price using keras.layers.recurrent.LSTM
- pcb: correlation analysis of copper and a downstream material price
You can visit the spiders project and run shfe to crawling copper future price from Shanghai Futures Exchange.
There is a main method in each python file so you can run it easily.
the following is a sample of predict result of the mlp method:
while the following is a sample of predict result of the lstm method with epochs=500:
- Yinwei Li
- wechat: coridc
- email: [email protected]
Don't hesitate to contact me on any topics about this project at your convenience.
When contributing to this repository, you can first discuss the change you wish to make via issue, email, or any other method with the owners of this repository.
This project is licensed under the GNU General Public License v3.0 License - see the LICENSE file for details.
I'd like hat tip to anyone who use the codes or send me any proposals of the project.