README.md

iTransformer for Multivariate Time Series Forecasting

This folder contains the reproductions of the iTransformers for Multivariate Time Series Forecasting (MTSF).

Dataset

Extensive challenging multivariate forecasting tasks are evaluated as the benchmark. We provide the download links: Google Drive or Tsinghua Cloud.

Scripts

In each folder named after the dataset, we provide the iTransformer experiments under four different prediction lengths as shown in the table above.

# iTransformer on the Traffic Dataset

bash ./scripts/multivariate_forecasting/Traffic/iTransformer.sh

To evaluate the model under other input/prediction lengths, feel free to change the seq_len and pred_len arguments:

# iTransformer on the Electricity Dataset, where 180 time steps are inputted as the observations, and the task is to predict the future 60 steps

python -u run.py \
  --is_training 1 \
  --root_path ./dataset/electricity/ \
  --data_path electricity.csv \
  --model_id ECL_180_60 \
  --model $model_name \
  --data custom \
  --features M \
  --seq_len 180 \
  --pred_len 60 \
  --e_layers 3 \
  --enc_in 321 \
  --dec_in 321 \
  --c_out 321 \
  --des 'Exp' \
  --d_model 512 \
  --d_ff 512 \
  --batch_size 16 \
  --learning_rate 0.0005 \
  --itr 1

Training on Custom Dataset

To train with your own time series dataset, you can try out the following steps:

1. Read through the Dataset_Custom class under the data_provider/data_loader folder, which provides the functionality to load and process time series files.
2. The file should be csv format with the first column containing the timestamp and the following columns containing the variates of time series.
3. Set data=custom and modify the enc_in, dec_in, c_out arguments according to your number of variates in the training script.