python scripts/split.py --data ./data/raw.csv --output ./data
scripts/split.py generate train.csv and test.csv which drop target labels from raw data. Basically, train.csv contains columns('Month, Day, Hour, Quarter, P1(DayOfWeek), Demand') where year is 2017.
python scripts/augmentation.py --train ./data/train.csv --output ./data/train-augmented.csv --repeat 8
scripts/augmentation.py generated train-augmented.csv for training.
python main.py --train ./data/train-augmented.csv --test ./data/test.csv --output ./results --input-size 32 --hidden-size 128
Main scripts create ./results directory which includes model.h5 and prediction.csv.
- hidden_size: 128
- input_size: 32
