Updated machine learning model

[jan-janssen]

No description provided.

[codecov]

Codecov Report

❌ Patch coverage is 97.22222% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 97.39%. Comparing base (aa9e6d0) to head (820e9df).

Files with missing lines	Patch %	Lines
gmailsorter/ml/encoding.py	93.33%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #461      +/-   ##
==========================================
- Coverage   97.50%   97.39%   -0.12%     
==========================================
  Files          35       35              
  Lines        1484     1498      +14     
==========================================
+ Hits         1447     1459      +12     
- Misses         37       39       +2     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[jan-janssen]

@copilot Improve test coverage by adding additional unit tests for the changes in this pull request

[Copilot]

@copilot Improve test coverage by adding additional unit tests for the changes in this pull request

Added 7 new unit tests in tests/test_ml.py (commit $(git -C /home/runner/work/gmailsorter/gmailsorter rev-parse --short HEAD)):

• test_train_random_forest_respects_max_depth_and_min_samples_leaf – verifies the new max_depth / min_samples_leaf params are applied to the fitted model
• test_train_random_forest_with_n_jobs – verifies n_jobs is passed through correctly
• test_fit_machine_learning_models_max_depth_and_min_samples_leaf – same for fit_machine_learning_models
• test_fit_machine_learning_models_max_workers – verifies max_workers maps to n_jobs on the fitted model
• test_get_predictions_single_output_label – covers the single-output squeeze normalisation path
• test_to_sparse_matrix_dense – covers the dense (numpy) code path in _to_sparse_matrix
• test_to_sparse_matrix_sparse – covers the sparse (scipy CSR) code path in _to_sparse_matrix

[jan-janssen]

pre-commit.ci autofix

[jan-janssen]

@copilot resolve the merge conflicts in this pull request

[Copilot]

@copilot resolve the merge conflicts in this pull request

Resolved in commit cf1835d. The three conflicted files were resolved as follows:

• gmailsorter/ml/model.py: kept the single multi-output RandomForestClassifier approach from this PR
• gmailsorter/ml/encoding.py: kept the sparse matrix approach in one_hot_encoding/_encoding_helper, and incorporated the label_prefix parameter that main added to encode_df_for_machine_learning
• gmailsorter/google/mail.py: removed the methods that main moved into AbstractMailBox
• gmailsorter/base/mail.py (auto-merged with old API): updated filter_messages_from_server and fit_machine_learning_model_to_database to use the new load_model/store_model API and the updated fit_machine_learning_models signature (max_depth, min_samples_leaf params)

All 59 tests pass.

[jan-janssen]

pre-commit.ci autofix

[jan-janssen]

@copilot Improve unit test coverage for the newly added functionality to maintain the test coverage rate of over 97%

[Copilot]

@copilot Improve unit test coverage for the newly added functionality to maintain the test coverage rate of over 97%

Added the following new tests across three test files (commit 2a0e8ef):

tests/test_mail_base.py (via _StubMailBox):

• test_fit_machine_learning_model_forwards_max_depth_and_min_samples_leaf – verifies max_depth and min_samples_leaf are forwarded to fit_machine_learning_models
• test_filter_messages_from_server_no_model_skips – covers the new early-return when load_model() returns None
• test_filter_messages_from_server_empty_df_skips – covers the no-op when download_emails_for_label returns an empty DataFrame
• test_filter_messages_from_server_with_model – covers the full encode → predict → move path

tests/test_google_integration_units.py:

• Adds max_depth/min_samples_leaf assertions to the existing test_fit_machine_learning_model_to_database
• test_filter_messages_from_server_no_model_skips – same early-return path via GoogleMailBase

tests/test_ml.py:

• test_store_model_user_isolation – verifies models for different user_ids are stored and loaded independently
• test_load_model_missing_returns_empty_features_when_features_exist – verifies load_model returns (None, [], feature_lst) when there is no model row but features exist