Data validation in CI pipeline in MLOps - Practice Problems & Coding Challenges

Challenge - 5 Problems

🧠 Conceptual

intermediate

Purpose of Data Validation in CI Pipeline

Why is data validation important in a Continuous Integration (CI) pipeline for machine learning projects?

A To speed up the model training by skipping data checks

B To reduce the size of the dataset by removing random samples

C To automatically deploy models without human review

D To ensure that the data used for training meets quality standards before model training starts

💡 Hint

Think about what could happen if bad data enters the training process.

Data validation in CI pipelines helps catch errors or inconsistencies in data early, preventing poor model performance or failures later.

💻 Command Output

intermediate

Output of Data Validation Step in CI

What is the expected output when a data validation step in a CI pipeline detects missing values in a critical feature?

MLOps

run_data_validation --input data.csv --check missing_values

A SyntaxError: invalid syntax in command.

B Validation passed: No missing values detected.

C Validation failed: Missing values found in column 'age'.

D Error: File not found 'data.csv'.

💡 Hint

Consider what happens if the validation finds a problem in the data.

The validation step reports failure with details about the missing values to prevent further pipeline execution.

❓ Configuration

advanced

Configuring Data Validation in a CI YAML Pipeline

Which YAML snippet correctly configures a data validation step in a CI pipeline that runs a Python script validate_data.py and fails the pipeline if validation fails?

A

steps:
  - name: Data Validation
    run: python validate_data.py

B

steps:
  - name: Data Validation
    run: python validate_data.py
    fail-on-error: true

C

steps:
  - name: Data Validation
    run: python validate_data.py
    continue-on-error: true

D

steps:
  - name: Data Validation
    run: python validate_data.py
    fail-fast: true

💡 Hint

By default, CI steps fail if the command exits with an error code.

Option A runs the script normally; if the script exits with a non-zero code, the pipeline fails. Other options use invalid or unsupported keys.

❓ Troubleshoot

advanced

Troubleshooting Data Validation Failure in CI

A CI pipeline data validation step fails with the error: KeyError: 'target_column'. What is the most likely cause?

A The dataset does not contain the required 'target_column' feature.

B The validation script has a syntax error in the code.

C The CI runner does not have Python installed.

D The pipeline configuration YAML is missing the data validation step.

💡 Hint

KeyError usually means a missing key in a dictionary or dataframe column.

The error indicates the script tried to access a column that does not exist in the input data, causing the failure.

🔀 Workflow

expert

Order of Steps in a CI Pipeline with Data Validation

What is the correct order of steps in a CI pipeline that includes data validation, model training, and deployment?

A 2,1,3

B 2,3,1

C 1,2,3

D 3,2,1

💡 Hint

Think about what must happen before training and deployment.

Data validation must happen first to ensure quality data, then training uses that data, and finally deployment happens after a successful model is trained.