- Download data
- Partition data
- Train models
- Evaluate models
./download_data.sh "data/raw_data"
python partition_data.py \
-raw_data_folder "data/raw_data" \
-labels_file_path "data/raw_data/labels.csv" \
-sampling_frequency 300 \
-test_set_size 528 \
-n_splits 10 \
-baseline_wander_removal_filter True \
-output_dir "data/datasets/10_partitions"
python train_model.py \
-model_name "model1" \
-data_dir "data/datasets/10_partitions" \
-test_set_path "data/datasets/10_partitions/test.tf" \
-n_records 5528 \
-n_test_set_records 528 \
-record_size 9000 \
-n_folds 5 \
-max_n_epochs 3 \
-initial_lr 0.001 \
-lr_reduction_factor 0.1 \
-early_stopping_patience 15 \
-lr_reducer_patience 5 \
-batch_size 128 \
-labels_file_path "data/raw_data/labels.csv" \
-class_weighting True
-weights_file_path "data/weights" \
-output_model_path "models" \
-fixed_fold 2 \
-static_lr False
python evaluate_models.py \
-models_paths "data/models/model1_fold_*.h5" "data/models/model2_fold_*.h5" "data/models/model3_fold_*.h5" \
-model_names "model1" "model2" "model3" \
-test_set_path "data/datasets/10_partitions/test.tf" \
-record_size 9000 \
-batch_size 528 \
-output_plot_dir "plots"
- Python 3.6.9
- Tensorflow 2.1.0
- Pandas 1.0.1
- Numpy 1.18.1