dvc.lock

schema: '2.0'
stages:
  data_ingestion:
    cmd: python src/pipeline/stage_01_data_ingestion.py
    deps:
    - path: params.yaml
      md5: 4cfa1f6904004bb95e43b9b5e4a6a18f
      size: 1007
    - path: src/pipeline/stage_01_data_ingestion.py
      md5: bc57183d7d3289433346b4466b58c9ea
      size: 646
    params:
      params.yaml:
        data_ingestion.local_data_file: artifacts/data_ingestion/data/data.csv
        data_ingestion.root_dir: artifacts/data_ingestion
        data_ingestion.source_URL: https://raw.githubusercontent.com/GregaVrbancic/Phishing-Dataset/master/dataset_full.csv
        data_ingestion.unzip_dir: artifacts/data_ingestion
    outs:
    - path: artifacts/data_ingestion/data/
      md5: 3fdb83a6e88e020c4db720e2e716f573.dir
      size: 25067193
      nfiles: 1
    - path: artifacts/data_ingestion/ingected_data/ingected_test_dir
      md5: eb26f17832ecc7565d6a93d54129bd4c.dir
      size: 5008202
      nfiles: 1
    - path: artifacts/data_ingestion/ingected_data/ingected_train_dir
      md5: e5539b3c8987a22b666b4b63fe9c424f.dir
      size: 20037507
      nfiles: 1
  data_validation:
    cmd: python src/pipeline/stage_02_data_validation.py
    deps:
    - path: params.yaml
      md5: 4cfa1f6904004bb95e43b9b5e4a6a18f
      size: 1007
    - path: src/components/data_validation.py
      md5: b69e66194aec698b6b23db562a5aa164
      size: 5631
    - path: src/pipeline/stage_02_data_validation.py
      md5: c1ff7e64b1d7065640fb058da503c46c
      size: 759
    params:
      params.yaml:
        data_validation.report_file_path: artifacts/data_validation/report.json
        data_validation.report_page_file_path: artifacts/data_validation/report.html
        data_validation.root_dir: artifacts/data_validation
        data_validation.schema_path_file: config/schema.yaml
    outs:
    - path: artifacts/data_validation/
      md5: 45278694a48a5db5de04672843820ea0.dir
      size: 57949790
      nfiles: 2
  data_preprocessing:
    cmd: python src/pipeline/stage_03_data_preprocessing.py
    deps:
    - path: params.yaml
      md5: 4cfa1f6904004bb95e43b9b5e4a6a18f
      size: 1007
    - path: src/components/data_preprocessing.py
      md5: ce987bc5e5fe1f518aa0d2be54bca4aa
      size: 3397
    - path: src/pipeline/stage_03_data_preprocessing.py
      md5: d2b41d338a98855cb8a5173a5665f37e
      size: 900
    params:
      params.yaml:
        data_preprocessing.cluster_model_path: models/kmean/kmean_model.pkl
        data_preprocessing.cluster_number_path: artifacts/data_preprocessing/cluster_number.json
        data_preprocessing.elbow_file_path: artifacts/data_preprocessing/kmeans_elbow.png
        data_preprocessing.root_dir: artifacts/data_preprocessing
    outs:
    - path: artifacts/data_preprocessing
      md5: 0f4f4b8bf843b4bf84ede772f134d6dc.dir
      size: 25765430
      nfiles: 3