API testing for data quality

It is the process of automatic data validation that is transmitted or returned via an API, with the aim of ensuring:

• Validity (for example, all emails have a valid format)

• Completeness (no empty mandatory fields)

• Consistency (the date format is always the same, IDs are unique)

• Accuracy (data matches the expected source of truth)

• Freshness (no outdated data, e.g., the last updated date is not older than X days)

What is being tested?

1. 1. Response schema validation - Does the API return all the required data in the correct format?

   2. Data integrity - Are the data correctly linked (e.g., the user has a valid address)?

   3. Data quality rules - Do the values adhere to business rules?

   4. Edge cases - What happens if some data is invalid, empty, or incomplete?

   5. Cross-source validation - Comparing the API responses with the database or other services.

Data Quality QA on AWS

1. 1. The data comes in S3 (CSV).

   2. Glue DataBrew It automatically triggers profiling.

   3. Lambda the function triggers Deequ validation.

   4. The results send an alert to CloudWatch + SNS (email/slack) If there is an error.

1. Data Splitting

• Train set - used for training the model.

• Validation set - used for hyperparameter tuning and preventing overfitting.

• Test set - used for final evaluation of the model's performance.

2. Most Common Metrics

Depends on the type of problem:

➤ Classification

• • Accuracy - the proportion of correctly predicted classes.

  • Precision, Recall, F1-score

  • ROC AUC - area under the ROC curve.

  • Confusion matrix - displays TN, TP, FN, FP.

Objectives:

• • Data loading and basic overview

  • Identification of missing values

  • Checking data types and duplicates

  • Outlier detection

  • Consistency check and data validation

1. Introduction to Artificial Intelligence (AI) for Testers

• What is AI and Why it Matters to QA Professionals

• Difference Between Traditional Software and AI Systems

• Examples where AI is used (chatbots, recommendations, vision, text classification, etc.)

2. What is AI testing?

• Definition and Goals of AI Testing

• Difference Between “Testing AI Systems” and “Using AI for Testing”

• Which components are tested in AI models (data, models, predictions, performance)

3. Key Challenges in AI Testing

• Non-functionality and non-linear behavior of the model

• Unpredictability and “black-box” nature

• Bias and unfair decisions made by the model

• Lack of deterministic expectations

4. Types of AI System Testing

• Data Testing (Data Quality)

• Model Testing (Accuracy, Precision)

• Model Behavior Testing (Boundary Testing)

• Evaluation of Metrics and Errors

5. Tools and Frameworks for AI Testing

• • Python + Pandas