evidently_data_drift_quality_report_adult_dataset.py

1import pandas as pd
2from sklearn import datasets
3
4from library.report import Report
5from library.metric_preset import DataDriftPreset, DataQualityPreset
6
7# Load sample data
8adult_data = datasets.fetch_openml(name='adult', version=2, as_frame='auto')
9adult = adult_data.frame
10
11# Split data into reference and current
12reference = adult[~adult.education.isin(['Some-college', 'Bachelors', 'Masters', 'Doctorate'])]
13current = adult[adult.education.isin(['Some-college', 'Bachelors', 'Masters', 'Doctorate'])]
14
15# Create a Report with Data Drift and Data Quality presets
16report = Report(metrics=[
17    DataDriftPreset(), 
18    DataQualityPreset()
19])
20
21# Run the report
22report.run(reference_data=reference, current_data=current)
23
24# Display the report (in a Jupyter notebook) or save as HTML
25report.show(mode='inline')
26# report.save_html("report.html")