TransformersEnergyTS

Key Directories and Modules

Entire Models Alongside our User-Friendly Wrappers for Training

• TransformersModels/AutoformerAPI.py
• TransformersModels/CrossformerAPI.py
• TransformersModels/FedformerAPI.py
• TransformersModels/InformerAPI.py
• TransformersModels/LogSparseAPI.py
• TransformersModels/TSMixerAPI.py

Synythetic Data Framework

• data_generator.py

Synthetic Data Framework Demo

• internal_demos/data_generator_demo.ipynb

Data Plots and Statistical Tests

• internal_demos/synth_analysis.ipynb

Wind Data Demo

• internal_demos/wind_data_demo.ipynb

Experiment Results

• results/

Our Synthetic Data (Extended from Liu et al. 2023, Pyraformer)

Forecasting Results

Multivariate

Univariate

---

For Transparency

Datasets Used in Experiments

• SYNTHDataset/SYNTHh1.csv
• SYNTHDataset/SYNTH_additive.csv
• SYNTHDataset/SYNTH_additive_reversal.csv
• SYNTHDataset/SYNTH_multiplicative.csv
• SYNTHDataset/SYNTH_multiplicative_reversal.csv
• WINDataset/DEWINDh_small.csv
• ETTDataset/ETTh1.csv

Model Training and Forecasting Logs

• new_forecast_notebook.ipynb

Results Analysis

• results_analysis.ipynb

(Outdated) Prototype

• prototype.ipynb

Note that results, data and code in the Prototype have since been reworked in the newer modules and directories.