Welcome to costcla’s documentation!

CostSensitiveClassification

costcla is a Python module for cost-sensitive machine learning (classification) built on top of Scikit-Learn, SciPy and distributed under the 3-Clause BSD license.

In particular, it provides:

1. A set of example-dependent cost-sensitive algorithms
2. Different reald-world example-dependent cost-sensitive datasets.

The project is part of the PhD research of Alejandro Correa Bahnsen.

Installation

You can install costcla with pip:

# pip install costcla

Documentation

Documentation is available at http://albahnsen.github.com/CostSensitiveClassification

Tutorials are available at http://albahnsen.com/CostSensitiveClassification/Tutorials.html

Development

Source code is available at https://github.com/albahnsen/CostSensitiveClassification

Contents:

• Introduction to Example-Dependent Cost-Sensitive Classification
  • Example-Dependent Cost-Sensitive Cost-Matrix
• Datasets
• Metrics
• Models
  • BayesMinimumRiskClassifier
  • ThresholdingOptimization
  • CostSensitiveLogisticRegression
  • CostSensitiveDecisionTreeClassifier
  • CostSensitiveRandomForestClassifier
  • CostSensitiveBaggingClassifier
  • CostSensitivePastingClassifier
  • CostSensitiveRandomPatchesClassifier
• Probcal
• Sampling
• CostCla Tutorials
  • Tutorial: Example-Dependent Cost-Sensitive Credit Scoring using CostCla
  • Slides: Example-Dependent Cost-Sensitive Fraud Detection using CostCla
• Change Log
  • Version 0.5 (2016-01-29)
  • Version 0.4 (2015-05-26)
  • Version 0.03 (2014-09-19)