Vaisshhh/data-quality-classifier

Data Quality Classifier

Overview

The Data Quality Classifier is a machine learning model developed as part of an automated data quality monitoring platform.

The model classifies datasets into quality categories based on quality-related features generated by the data quality pipeline.

Model Details

• Model: DecisionTreeClassifier
• Framework: PyCaret
• Target: quality
• Training rows: 200
• Training columns: 6
• Model selection: PyCaret compare_models()
• Experiment tracking: MLflow

Quality Classes

The model predicts the following quality categories:

• Excellent
• Good
• Moderate
• Poor

Performance

The selected model achieved the following evaluation results:

Metric	Score
Accuracy	85.00%
Precision	85.64%
Recall	85.00%
F1 Score	85.07%

Training and Experiment Tracking

PyCaret was used to compare classification models and select the best-performing model.

MLflow is used to track:

• Training parameters
• Model metrics
• Model artifacts
• Model comparison results
• Prediction distributions
• Prediction drift between runs

Intended Use

This model is intended for automated classification of dataset quality within a data quality monitoring pipeline.

It can be used alongside data profiling, validation, quality scoring, and dashboard reporting components.

Limitations

• Model performance depends on the quality and representativeness of the training data.
• The current training dataset contains 200 rows.
• Predictions should be interpreted together with the underlying data quality metrics.
• The model should be retrained when the characteristics of incoming datasets change significantly.

Project Components

The overall platform includes:

• Dataset ingestion
• Dataset profiling
• Great Expectations validation
• Data quality scoring
• Machine learning classification
• MLflow experiment tracking
• Prediction drift monitoring
• Dashboard reporting

Files

• dataset_quality_model.pkl - Trained classification model
• model_metrics.csv - Model evaluation metrics
• model_comparison_results.csv - PyCaret model comparison results
• requirements.txt - Python dependencies