The project is designed to tackle time series. It also includes various techniques of data exploration, data processing, dimensionality reduction, feature engineering and cross validation for time series.
This project compares machine learning models in order to learn how to work with them and to understand the ratio of effort to the end result of each one.
In the development process, the following models have been used for prediction:
- Moving average
- (S)ARIMA model
- Linear regression
- Lasso regression (L1)
- Ridge regression (L2)
- XGBoost
- Prophet model
- LSTM model
The goal of the project was to understand the workflow behind time-series processing as well as hands-on experience with various machine learning models.