Setup
git clone https://github.com/phamdinhdat-ai/Time_series_projectcd Time_series_projectgit checkout SPP_2023_v1pip install -r requirement.txt
Data
- In
/datafolder, create 4 sub-folder: static: add static dataset to this folderdynamic: add dynamic dataset to this folder5_fold_dynamic: add k-fold dynamic dataset5_fold_static: add k-fold static dataset
Parameter Training
--model_type: type of network to training--data_type: type of dataset--num_classes: numbers of class in dataset--epochs: numbers of epoch--batch_size: numbers of batch--squence_lenght: lenght of the sliding window--overlap: percent of overlap window
Run Code
-
Training data with two option datasets: static and dynamic -
For dynamic dataset:
python .\main.py --model_type mlp --data_type dynamic --num_classes 4 --epochs 100 --batch_size 256 --sequence_lenght 10 --overlap 0.1 -
For static dataset:
python .\main.py --model_type mlp --data_type static --num_classes 13 --epochs 100 --batch_size 256 --sequence_lenght 10 --overlap 0.1 -
For k-fold dynamic dataset:
!python main_k_fold.py --model_type mlp --data_type k_fold_dynamic --num_classes 4 --epochs 100 --batch_size 512 --sequence_lenght 10 --overlap 0.9 -
For k-fold static dataset:
!python main_k_fold.py --model_type mlp --data_type k_fold_static --num_classes 13 --epochs 100 --batch_size 512 --sequence_lenght 10 --overlap 0.9