This project aims to provide 5-minutes short-term price prediction based on market data and sentiment data.
First of all, run cells in EDA.ipynb
, which conducts some exploratory data analysis and generates some plots. It gives user a clear idea about how the model is performing.
Then, run
python pipeline.py regularization_search
in the terminal. The pipeline.py
conducts the hyperparameter search based on the information in searches.py
. This framework is expandable: user could try edit the code in get_rolling_data
function in pipeline.py
and define new hyperparameter search in pipeline.py
to conduct other kinds of hyperparameter search.
Then open the notebook ResultAnalysis.ipynb
to conduct the result analysis, select the best combinations, and visualize.