Skip to main content

Python/Pyspark Package for Model-agnostic Evaluation and Diagnosis.

Project description

logo

Pytalite: Model-agnostic diagnostics and evaluation tool for Python and Pyspark

Overview

Evaluating model performance and diagnosing models are important steps in machine learning model development. Pytalite provides a straightforward wrapper for visualizing model performance and diagnosing machine learning models. It enables users to visualize key snapshots of model performance, e.g. discrete precision-recall curves, probability density plots, model-agnostic feature importance, partial dependence plots, accumulative local effect plots, and feature correlation plots.

Requirements

Pytalite for python is developed under python 3.7, but is compatible with python 2.7. Pytalite for pyspark is developed to support spark 2.0 and above.

matplotlib ≥ 2.2.x (although 1.4.3 also works, higher version is recommended)

numpy ≥ 1.9.x

scipy ≥ 0.15.x

multiprocess ≥ 0.70.4

Package Structure

Highlights

Pytalite provides the following model evaluation and diagnostic algorithms:

  • Discrete Precision/Recall Plot (Binary-Classification only)
  • Feature Correlation Plot (Binary-Classification only)
  • Probability Density Plot (Binary-Classification only)
  • Feature Importance Plot (Binary-Classification only)
  • Accumulated Local Effect Plot (Binary-Classification / Regression, numerical feature)
  • Partial Dependence Plot (Binary-Classification / Regression)

Examples

See examples folder for usage examples.

Gallery

Discrete Precision/Recall Plot


Feature Importance Plot


Feature ALE Plot for numerical features

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

pytalite-1.0.0.tar.gz (27.1 kB view hashes)

Uploaded Source

Built Distribution

pytalite-1.0.0-py3-none-any.whl (29.1 kB view hashes)

Uploaded Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page