Modern Time Series Analysis

MTSA (Modern Time Series Analysis) is a library dedicated to the field of time series forecasting. Our primary objective is to provide a comprehensive collection of both classical and deep learning-based algorithms for tackling time series forecasting tasks.

We will gradually enhance and expand our library as the TSA (Time Series Analysis) course progresses.

TSA home page

Usage examples

python main.py --data_path ./dataset/ETT/ETTh1.csv --dataset ETT --target OT --model MeanForecast

python main.py --data_path ./dataset/ETT/ETTh1.csv --dataset ETT --target OT --model TsfKNN --n_neighbors 1 --msas MIMO --distance euclidean

Datasets

All datasets can be found here.

• M4
• ETT
• Traffic
• Electricity
• Exchange-Rate
• Weather
• ILI(illness)

Models

• ZeroForecast
• MeanForecast
• TsfKNN
• LinearRegressionForecast
• ExponentialSmoothingForecast

Transformations

• IdentityTransform
• Normalization
• Standardization
• Mean Normalization
• Box-Cox

Metrics

• MSE
• MAE
• MASE
• MAPE
• SMAPE