About

The Data-Analysis Metrics Library, or DataEval, focuses on characterizing image data and its impact on model performance across classification and object-detection tasks.

Data Analytics

The scope of data analytics is very broad, but as DataEval is developed, we look for impactful metrics such as the ones below as a guide towards feature development.

Model-agnostic metrics that bound real-world performance

• relevance/completeness/coverage

• metafeatures (data complexity)

Model-specific metrics that guide model selection and training

• dataset sufficiency

• data/model complexity mismatch

Metrics for post-deployment monitoring of data with bounds on model performance to guide retraining

• dataset-shift metrics

• model performance bounds under covariate shift

• guidance on sampling to assess model error and model retraining