This project uses lightgbm model for predicting the future prices of cryptocurrencies using historical data to provide insights into potential trends in prices and uses dataset provided as part of the G-Research Crypto Forecasting Challenge
-
Programming language used: Python
-
Libraries such as Matplotlib, SKLearn, Pandas, NumPy
-
SciPy(pearsonr)
-
LightGBM(LGBM REGRESSOR)
-
Data Preprocessing: handling the missing values and outliers
-
Feature Engineering: creating relevant technical indicators to be used for training the model
-
Model: LightGBM
-
Training and Testing: use of k-fold cross validation technique for training the model
-
The model can be trained on CPU but a minimum RAM of 16GB is required.
-
Run the code on any Python IDE or use the Jupyter notebook support for Visual Studio Code
-
Download the complete dataset here: G-Research Crypto Forecasting Challenge