# SPDX-License-Identifier: Apache-2.0
_target_: data.datasets.create_datasets
config:
    source_path: /workspace/datasets/pems_bay/pems_bay.csv
    dest_path: /workspace/datasets/pems_bay/
    binarized: False
    graph: graph.bin
    graph_partitions: 1
    partition_joining_coef: 1
    train_range:
      - '2017-1-1'
      - '2017-5-8'
    valid_range:
      - '2017-5-8'
      - [2017,5,25,17,50]
    test_range:
      - [2017,5,25,17,50]
      - '2017-7-01'
    dataset_stride: 1
    scale_per_id: True
    encoder_length: 12
    input_length: 12
    example_length: 24
    MultiID: True
    features:
      - name: 'id'
        feature_type: 'ID'
        feature_embed_type: 'CATEGORICAL'
        cardinality: 325
      - name: 'Timestamp'
        feature_type: 'TIME'
        feature_embed_type: 'DATE'
      - name: 'Avg Speed'
        feature_type: 'TARGET'
        feature_embed_type: 'CONTINUOUS'
        scaler:
            _target_: sklearn.preprocessing.StandardScaler
      - name: 'Station'
        feature_type: 'STATIC'
        feature_embed_type: 'CATEGORICAL'
        cardinality: 326
      - name: 'Freeway #'
        feature_type: 'STATIC'
        feature_embed_type: 'CATEGORICAL'
        cardinality: 9
      - name: 'Direction of Travel'
        feature_type: 'STATIC'
        feature_embed_type: 'CATEGORICAL'
        cardinality: 5
      #- name: 'Station Length'
      #  feature_type: 'STATIC'
      #  feature_embed_type: 'CONTINUOUS'
      #  scaler:
      #      _target_: sklearn.preprocessing.StandardScaler
      - name: 'Avg Occupancy'
        feature_type: 'OBSERVED'
        feature_embed_type: 'CONTINUOUS'
        scaler:
            _target_: sklearn.preprocessing.StandardScaler
      - name: 'Total Flow' 
        feature_type: 'OBSERVED'
        feature_embed_type: 'CONTINUOUS'
        scaler:
            _target_: sklearn.preprocessing.StandardScaler
      - name: 'Day of week'
        feature_type: 'KNOWN'
        feature_embed_type: 'CATEGORICAL'
        cardinality: 8
      - name: 'Month'
        feature_type: 'KNOWN'
        feature_embed_type: 'CONTINUOUS'
        scaler:
            _target_: sklearn.preprocessing.StandardScaler
      - name: 'Day'
        feature_type: 'KNOWN'
        feature_embed_type: 'CONTINUOUS'
        scaler:
            _target_: sklearn.preprocessing.StandardScaler
      - name: 'Hour'
        feature_type: 'KNOWN'
        feature_embed_type: 'CONTINUOUS'
        scaler:
            _target_: sklearn.preprocessing.StandardScaler
      - name: 'Minute'
        feature_type: 'KNOWN'
        feature_embed_type: 'CONTINUOUS'
        scaler:
            _target_: sklearn.preprocessing.StandardScaler

    time_series_count: 325