titanic_dvc/dvc.yaml

stages:
  make_dataset:
    desc: Download data from Kaggle, create data dictionary and summary dtable
    cmd: python3 src/data/make_dataset.py -c titanic -tr train.csv -te test.csv -o
      ./data/raw
    deps:
    - src/data/make_dataset.py
    params:
    - dtypes
    outs:
    - data/raw/test.csv
    - data/raw/train.csv
    - reports/figures/data_dictionary.tex
    - reports/figures/table_one.tex
  encode_labels:
    desc: Convert categorical labels to integer values and save mapping
    cmd: python3 src/data/encode_labels.py -tr data/raw/train.csv -te data/raw/test.csv
      -o data/interim
    deps:
    - data/raw/test.csv
    - data/raw/train.csv
    - src/data/encode_labels.py
    params:
    - dtypes
    outs:
    - data/interim/label_encoding.yaml
    - data/interim/test_categorized.csv
    - data/interim/train_categorized.csv
  impute_nan:
    desc: Replace missing values for age with mean values from training dataset.
    cmd: python3 src/data/replace_nan.py -tr data/interim/train_categorized.csv -te
      data/interim/test_categorized.csv -o data/interim
    deps:
    - data/interim/test_categorized.csv
    - data/interim/train_categorized.csv
    - src/data/replace_nan.py
    params:
    - imputation
    outs:
    - data/interim/test_nan_imputed.csv
    - data/interim/train_nan_imputed.csv
  build_features:
    desc: Optional feature engineering and dimensionality reduction
    cmd: python3 src/features/build_features.py -tr data/interim/train_nan_imputed.csv
      -te data/interim/test_nan_imputed.csv -o data/interim/
    deps:
    - data/interim/test_nan_imputed.csv
    - data/interim/train_nan_imputed.csv
    - src/features/build_features.py
    params:
    - feature_eng
    - random_seed
    outs:
    - data/interim/test_featurized.csv
    - data/interim/train_featurized.csv
  normalize_data:
    desc: Optionally normalize features by fitting transforms on the training dataset.
    cmd: python3 src/features/normalize.py -tr data/interim/train_featurized.csv -te
      data/interim/test_featurized.csv -o data/processed/
    deps:
    - data/interim/test_featurized.csv
    - data/interim/train_featurized.csv
    - src/features/normalize.py
    params:
    - normalize
    outs:
    - data/processed/test_processed.csv
    - data/processed/train_processed.csv
  split_train_dev:
    desc: Split training data into the train and dev sets using stratified K-fold
      cross validation.
    cmd: python3 src/data/split_train_dev.py -tr data/processed/train_processed.csv
      -o data/processed/
    deps:
    - data/processed/train_processed.csv
    - src/data/split_train_dev.py
    params:
    - random_seed
    - train_test_split
    outs:
    - data/processed/split_train_dev.csv
  train_model:
    desc: Train the specified classifier using the pre-allocated stratified K-fold
      cross validation splits and the current params.yaml settings. Track metrics
      with Git
    cmd: python3 src/models/train_model.py -tr data/processed/train_processed.csv
      -cv data/processed/split_train_dev.csv
    deps:
    - data/processed/split_train_dev.csv
    - data/processed/train_processed.csv
    - src/models/train_model.py
    params:
    - classifier
    - model_params
    - random_seed
    - train_test_split.target_class
    outs:
    - models/estimator.pkl
    metrics:
    - results/metrics.json:
        cache: false
  predict_output:
    desc: Predict output on held-out test set for submission to Kaggle.
    cmd: python3 src/models/predict.py -te data/processed/test_processed.csv -rd results/
      -md models/
    deps:
    - data/processed/test_processed.csv
    - models/estimator.pkl
    - src/models/metrics.py
    - src/models/predict.py
    params:
    - predict
    - train_test_split.target_class
    outs:
    - results/test_predict_binary.csv
    - results/test_predict_proba.csv