exp w/ max_features=10, sources redacted

.github/workflows/cml.yaml

@@ -3,10 +3,12 @@ on: pull_request
 jobs:
   run:
     runs-on: [ubuntu-latest]
+    container: docker://dvcorg/cml-py3:latest
     steps:
-      - uses: iterative/setup-cml@v2
-      - uses: iterative/setup-dvc@v1
-      - uses: actions/checkout@v3
+      # - uses: iterative/setup-cml@v2
+      # - uses: iterative/setup-dvc@v1
+      - uses: actions/checkout@v2
+      - name: CML run
         with:
           fetch-depth: 2
       # Needed for https://github.com/iterative/example-repos-dev/issues/225
@@ -24,7 +26,7 @@ jobs:
             git fetch origin main:main
           fi
 
-          dvc pull eval
+          dvc repro
           dvc plots diff $PREVIOUS_REF workspace \
             --show-vega --targets ROC | json5 > vega.json
           vl2svg vega.json roc.svg

dvc.lock

@@ -28,27 +28,27 @@ stages:
       nfiles: 2
     - path: src/prepare/collate.py
       hash: md5
-      md5: 196f211d4aa609e4623bfbaa78048d6d
+      md5: 075f9c5c4e5e0a24a7baece2c1dffe62
       size: 1328
     params:
       params.yaml:
         prepare.seed: 42
     outs:
     - path: data/collated/data.csv
       hash: md5
-      md5: 6c011536ae995e2558840b7064599104
+      md5: a8ee15642e77d9eb17c55b58cf0d9dbc
       size: 92520291
   split:
-    cmd: python src/prepare/split.py data/data.csv
+    cmd: python src/prepare/split.py data/collated/data.csv
     deps:
-    - path: data/data.csv
+    - path: data/collated/data.csv
       hash: md5
       md5: a8ee15642e77d9eb17c55b58cf0d9dbc
       size: 92520291
     - path: src/prepare/split.py
       hash: md5
-      md5: 7b17f69ba203ff9290e23cccba4330dd
-      size: 1794
+      md5: a543b0bfb12d98b8b92faf6c01b3102f
+      size: 1607
     params:
       params.yaml:
         prepare.seed: 42
@@ -73,21 +73,21 @@ stages:
       size: 4065
     params:
       params.yaml:
-        featurize.max_features: 200
+        featurize.max_features: 10
         featurize.ngrams: 2
     outs:
     - path: data/features
       hash: md5
-      md5: c0fde86f821aa3994f8699dc6dd1923b.dir
-      size: 15293681
+      md5: 145b99ce99000a8fc8be0e08c3517578.dir
+      size: 2882605
       nfiles: 2
   train:
     cmd: python src/train.py data/features model.pkl
     deps:
     - path: data/features
       hash: md5
-      md5: c0fde86f821aa3994f8699dc6dd1923b.dir
-      size: 15293681
+      md5: 145b99ce99000a8fc8be0e08c3517578.dir
+      size: 2882605
       nfiles: 2
     - path: src/train.py
       hash: md5
@@ -101,29 +101,29 @@ stages:
     outs:
     - path: model.pkl
       hash: md5
-      md5: cb60759246d8999b8d457ba56607d5d6
-      size: 2278913
+      md5: 8a6cd2f4570d1170fbdb4883b82e92dc
+      size: 1322189
   evaluate:
     cmd: python src/evaluate.py model.pkl data/features
     deps:
     - path: data/features
       hash: md5
-      md5: c0fde86f821aa3994f8699dc6dd1923b.dir
-      size: 15293681
+      md5: 145b99ce99000a8fc8be0e08c3517578.dir
+      size: 2882605
       nfiles: 2
     - path: model.pkl
       hash: md5
-      md5: cb60759246d8999b8d457ba56607d5d6
-      size: 2278913
+      md5: 8a6cd2f4570d1170fbdb4883b82e92dc
+      size: 1322189
     - path: src/evaluate.py
       hash: md5
       md5: a1a59f55636170fb56e0c6afd3e28fa4
       size: 3315
     outs:
     - path: eval
       hash: md5
-      md5: 1332ef2ad9a41a6d73681be5ea93c15b.dir
-      size: 6196803
+      md5: cbbbc5e755b9bc6ac69c65213c4e4a51.dir
+      size: 6116699
       nfiles: 8
   redact:
     cmd: python src/prepare/redact.py data/cleaned

dvc.yaml

@@ -12,35 +12,56 @@ artifacts:
       - classification
       - news
 stages:
+
+  # sources not redacted
+
   # clean:
   #   cmd: python src/prepare/clean.py data/raw
   #   deps:
   #     - data/raw
   #     - src/prepare/clean.py
   #   outs:
   #     - data/cleaned
-  # redact:
-  #   cmd: python src/prepare/redact.py data/cleaned
-  #   deps:
-  #     - data/cleaned
-  #     - src/prepare/redact.py
-  #   outs:
-  #     - data/redacted
   # collate:
-  #   cmd: python src/prepare/collate.py data/redacted
+  #   cmd: python src/prepare/collate.py data/cleaned
   #   deps:
-  #     - data/redacted
+  #     - data/cleaned
   #     - src/prepare/collate.py
   #   params:
   #     - prepare.seed
   #   outs:
   #     - data/collated/data.csv
+
+  # sources redacted
+
+  clean:
+    cmd: python src/prepare/clean.py data/raw
+    deps:
+      - data/raw
+      - src/prepare/clean.py
+    outs:
+      - data/cleaned
+  redact:
+    cmd: python src/prepare/redact.py data/cleaned
+    deps:
+      - data/cleaned
+      - src/prepare/redact.py
+    outs:
+      - data/redacted
+  collate:
+    cmd: python src/prepare/collate.py data/redacted
+    deps:
+      - data/redacted
+      - src/prepare/collate.py
+    params:
+      - prepare.seed
+    outs:
+      - data/collated/data.csv
+
   split:
-    # cmd: python src/prepare/split.py data/collated/data.csv
-    cmd: python src/prepare/split.py data/data.csv
+    cmd: python src/prepare/split.py data/collated/data.csv
     deps:
-      # - data/collated/data.csv
-      - data/data.csv
+      - data/collated/data.csv
       - src/prepare/split.py
     params:
       - prepare.seed

params.yaml

@@ -3,7 +3,7 @@ prepare:
   seed: 42
 
 featurize:
-  max_features: 200
+  max_features: 10
   ngrams: 2
 
 train:

src/prepare/split.py

@@ -25,11 +25,8 @@ def split(dataset, train_out, test_out, split, seed):
         X, y, test_size=split, stratify=y, random_state=seed
     )
 
-    # train = pd.concat([X_train, y_train], axis=0)
-    # test = pd.concat([X_test, y_test], axis=0)
-
-    train = pd.DataFrame({"id": X_train['id'], "text": X_train['text'], "label": y_train})
-    test = pd.DataFrame({"id": X_test['id'], "text": X_test['text'], "label": y_test})
+    train = pd.concat([X_train, y_train], axis=1)
+    test = pd.concat([X_test, y_test], axis=1)
 
     # save train and test data
     os.makedirs(os.path.join("data", "prepared"), exist_ok=True)