CONTIME

Addressing prediction delays in time series forecasting: A continuous GRU Approach with derivative regularization Newly accepted KDD 2024

OTHER DATASETS WILL BE UPLOADED SOON!! (SORRY FOR THE LATE UPLOAD)

Installation

pip install contime_kdd
conda activate contime_kdd

Training code

python contime.py --dataset 'AWS' --batch 256 --task 'forecasting' --epoch 100 --model 'contime' --seq_len 104 --pred_len 24 --stride_len 1 --alpha 0.8  --lr 0.005 --beta 0.1 --seed 2021 --training 'True' --note '0126' --missing_rate 0 --data_name 'AMZN'

Testing code

python contime.py --dataset 'AWS' --batch 256 --task 'forecasting' --epoch 100 --model 'contime' --seq_len 104 --pred_len 24 --stride_len 1 --alpha 0.8  --lr 0.005 --beta 0.1 --seed 2021 --training '' --note '0126' --missing_rate 0 --data_name 'AMZN'