All Tools
N
MonitoringFreemiumOpen Source
NANNYML
Post-deployment ML monitoring without ground truth labels
Apache-2.0
ABOUT
Monitoring ML model performance in production typically requires ground truth labels, which are often delayed by days or weeks — meaning model degradation goes undetected until it is too late. NannyML solves this with confidence-based performance estimation (CBPE) that estimates model performance without needing ground truth, combined with data drift detection that alerts on distribution shifts the moment they occur.
INSTALL
pip install nannymlINTEGRATION GUIDE
1. Monitor production ML model accuracy in real time without waiting for ground truth labels to arrive
2. Alert on data drift and model degradation automatically with confidence-based performance estimates
3. Implement a proactive ML monitoring system that detects issues before they impact business metrics
4. Track model performance across segments to identify which customer groups are affected by drift
5. Build a monitoring dashboard that combines drift detection with estimated performance metrics for ops teams
TAGS
model-monitoringdrift-detectionperformance-estimationmlopspythondata-sciencepost-deployment