CI/CD for ML pipelines in ML Python - Model Metrics & Evaluation

Metrics & Evaluation - CI/CD for ML pipelines

Which metric matters for CI/CD in ML pipelines and WHY

In CI/CD for ML pipelines, the key metrics are model performance metrics like accuracy, precision, recall, and F1 score, plus pipeline reliability metrics such as build success rate and deployment frequency. These metrics matter because they show if the model is improving and if the pipeline runs smoothly without errors. Monitoring these helps catch problems early and keep the ML system working well.

Confusion matrix example for model evaluation in CI/CD

      Actual \ Predicted | Positive | Negative
      -------------------|----------|---------
      Positive           |    80    |   20
      Negative           |    10    |   90
    

This confusion matrix helps calculate precision, recall, and accuracy to decide if the model is good enough to deploy.

Precision vs Recall tradeoff in ML pipeline deployment

Imagine a spam filter model in your pipeline:

• High precision means few good emails are wrongly marked as spam. This avoids annoying users.
• High recall means most spam emails are caught. This keeps inboxes clean.

Depending on your goal, you might prioritize one over the other before deploying the model.

Good vs Bad metric values for CI/CD in ML pipelines

• Good: Accuracy above 90%, precision and recall balanced above 85%, pipeline build success rate near 100%, fast deployment times.
• Bad: Accuracy below 70%, precision or recall very low (below 50%), frequent pipeline failures, slow or manual deployments.

Common pitfalls in metrics for ML CI/CD pipelines

• Accuracy paradox: High accuracy but poor recall on rare classes can hide problems.
• Data leakage: Training on future data inflates metrics, causing bad deployments.
• Overfitting: Great training metrics but poor real-world results mean the model won’t generalize.
• Ignoring pipeline failures: Deploying models without checking pipeline health can cause downtime.

Self-check question

Your ML pipeline shows 98% accuracy but only 12% recall on fraud detection. Is this model good for production? Why or why not?

Answer: No, it is not good. The low recall means the model misses most fraud cases, which is critical to catch. High accuracy is misleading because fraud is rare, so the model mostly predicts non-fraud correctly but fails where it matters.

Key Result

In CI/CD for ML pipelines, balancing model performance metrics like precision and recall with pipeline reliability metrics ensures safe and effective deployments.

Practice

1. What is the main purpose of a CI/CD pipeline in machine learning projects?

easy

A. To manually update model parameters

B. To automate model training, testing, and deployment

C. To store large datasets only

D. To write machine learning algorithms from scratch

Solution

1. Step 1: Understand CI/CD pipeline role

   CI/CD pipelines automate repetitive tasks like training, testing, and deploying ML models.

2. Step 2: Compare options

   Only To automate model training, testing, and deployment describes automation of these tasks; others describe unrelated activities.

3. Final Answer:

   To automate model training, testing, and deployment -> Option B

4. Quick Check:

   CI/CD automates ML workflows = B [OK]

Hint: CI/CD means automating ML workflow steps [OK]

Common Mistakes:

• Thinking CI/CD is only for code storage
• Confusing manual updates with automation
• Assuming CI/CD writes ML algorithms

2. Which of the following is the correct YAML syntax snippet to trigger a CI/CD pipeline on every code push in GitHub Actions?

easy

A. on: [push]

B. trigger: push

C. when: code_push

D. event: commit

Solution

1. Step 1: Identify GitHub Actions trigger syntax

   GitHub Actions uses 'on:' keyword with event names like 'push' to trigger workflows.

2. Step 2: Validate options

   Only 'on: [push]' is valid YAML syntax for triggering on code push.

3. Final Answer:

   on: [push] -> Option A

4. Quick Check:

   GitHub Actions trigger = on: [push] [OK]

Hint: GitHub Actions uses 'on:' for triggers [OK]

Common Mistakes:

• Using 'trigger' instead of 'on'
• Wrong keywords like 'when' or 'event'
• Incorrect YAML list syntax

3. Given this simplified ML pipeline step in a CI/CD YAML file:

steps:
  - name: Train model
    run: python train.py --epochs 5
  - name: Test model
    run: python test.py
  - name: Deploy model
    run: python deploy.py

What will happen if train.py fails during execution?

medium

A. The pipeline stops and does not run test or deploy steps

B. The pipeline continues to test and deploy steps regardless

C. Only the deploy step is skipped, test runs

D. The pipeline retries the train step automatically

Solution

1. Step 1: Understand default CI/CD step behavior

   By default, if a step fails, the pipeline stops and next steps do not run.

2. Step 2: Apply to given steps

   If 'train.py' fails, 'test.py' and 'deploy.py' steps will not execute.

3. Final Answer:

   The pipeline stops and does not run test or deploy steps -> Option A

4. Quick Check:

   Step failure stops pipeline = A [OK]

Hint: Failed step stops pipeline unless configured otherwise [OK]

Common Mistakes:

• Assuming pipeline continues after failure
• Thinking only deploy step is skipped
• Believing automatic retries happen by default

4. You wrote this GitHub Actions step to deploy your ML model:

- name: Deploy model
  run: python deploy.py
  if: success()

But the deploy step runs even if the test step fails. What is the likely cause?

medium

A. The deploy.py script always succeeds regardless of tests

B. The 'if' condition should be 'if: failure()' to run on failure

C. GitHub Actions ignores 'if' conditions on run steps

D. The 'if: success()' condition is incorrectly indented

Solution

1. Step 1: Check YAML indentation rules

   In GitHub Actions, 'if:' must be at the same level as 'run:' to apply correctly.

2. Step 2: Analyze given snippet

   'if: success()' is indented under 'run:', so it is ignored and deploy runs always.

3. Final Answer:

   The 'if: success()' condition is incorrectly indented -> Option D

4. Quick Check:

   YAML indentation affects condition execution = D [OK]

Hint: Check YAML indentation for 'if:' conditions [OK]

Common Mistakes:

• Misplacing 'if:' under 'run:'
• Confusing success() with failure()
• Assuming scripts control step execution

5. You want to automate retraining your ML model weekly and also on every code push. Which CI/CD configuration best achieves this?

hard

A. trigger: push\nschedule: weekly\ncron: '0 0 * * 0'

B. on: push\nschedule: 'weekly' # invalid syntax

C. on: [push, schedule]\nschedule:\n - cron: '0 0 * * 0' # every Sunday midnight

D. on: [push]\nschedule: every Sunday

Solution

1. Step 1: Understand GitHub Actions triggers

   To trigger on code push and schedule, use 'on:' with both 'push' and 'schedule' keys.

2. Step 2: Validate cron syntax

   on: [push, schedule]\nschedule:\n - cron: '0 0 * * 0' # every Sunday midnight uses correct cron format for weekly schedule and valid YAML syntax.

3. Final Answer:

   on: [push, schedule]\nschedule:\n - cron: '0 0 * * 0' # every Sunday midnight -> Option C

4. Quick Check:

   Use 'on:' with push and schedule + valid cron [OK]

Hint: Use 'on:' with push and schedule plus cron for weekly [OK]

Common Mistakes:

• Using invalid schedule syntax
• Misplacing cron outside schedule
• Using unsupported keywords like 'trigger'