dvc.lock

schema: '2.0'
stages:
  prepare:
    cmd: python 00_data_preparation.py
    deps:
    - path: 00_data_preparation.py
      hash: md5
      md5: 79075a369d39e2d8ffcf88e74bb03776
      size: 1262
    - path: data/Train.csv
      hash: md5
      md5: 6117a2a9213e0159860c21e2516db548
      size: 5394818
    outs:
    - path: data/dataset/X_test.parquet
      hash: md5
      md5: 7c1389dcd5d85f1f13460ecbc8b21c9f
      size: 173169
    - path: data/dataset/X_train.parquet
      hash: md5
      md5: 0f518c1296f8b60d6f63c51907a24da6
      size: 346441
    - path: data/dataset/X_val.parquet
      hash: md5
      md5: e4acdfccf3daaab2fe937b82a0f43741
      size: 136839
    - path: data/dataset/y_test.parquet
      hash: md5
      md5: c3a1ed394f1b8823f40feca313a3248c
      size: 53693
    - path: data/dataset/y_train.parquet
      hash: md5
      md5: 0e910ab3ac73927c80d0d78c1a74b564
      size: 120136
    - path: data/dataset/y_val.parquet
      hash: md5
      md5: 4862803649e0edddd7c23da789f4e810
      size: 40063
  train:
    cmd: python 01_training.py
    deps:
    - path: 01_training.py
      hash: md5
      md5: 93e4afd970de70f126ae4fa94e2e1198
      size: 988
    - path: data/dataset/X_train.parquet
      hash: md5
      md5: 0f518c1296f8b60d6f63c51907a24da6
      size: 346441
    - path: data/dataset/X_val.parquet
      hash: md5
      md5: e4acdfccf3daaab2fe937b82a0f43741
      size: 136839
    - path: data/dataset/y_train.parquet
      hash: md5
      md5: 0e910ab3ac73927c80d0d78c1a74b564
      size: 120136
    - path: data/dataset/y_val.parquet
      hash: md5
      md5: 4862803649e0edddd7c23da789f4e810
      size: 40063
    params:
      params.yaml:
        colsample_bytree: 0.5899891961399993
        early_stopping_rounds: 15
        learning_rate: 0.06519967442047679
        max_depth: 11
        min_child_weight: 3
        n_estimators: 183
        subsample: 0.9526677078382316
    outs:
    - path: model.json
      hash: md5
      md5: b7c632fdb03aa4e3358a60f626a4e7ca
      size: 2413233
  evaluate:
    cmd: python 02_evaluation.py
    deps:
    - path: 02_evaluation.py
      hash: md5
      md5: bbc7ff5a52bb7bee57368cd1ebe964cb
      size: 2392
    - path: data/dataset/X_test.parquet
      hash: md5
      md5: 7c1389dcd5d85f1f13460ecbc8b21c9f
      size: 173169
    - path: data/dataset/y_test.parquet
      hash: md5
      md5: c3a1ed394f1b8823f40feca313a3248c
      size: 53693
    - path: model.json
      hash: md5
      md5: b7c632fdb03aa4e3358a60f626a4e7ca
      size: 2413233