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 modelmodel_metrics.csv - Model evaluation metricsmodel_comparison_results.csv - PyCaret model comparison resultsrequirements.txt - Python dependencies
