Skip to main content

scripts for calculating likelihood ratios

Project description

LIR Python Likelihood Ratio Library

This library provides a collection of scripts to aid calibration, and calculation and evaluation of Likelihood Ratios.

A simple score-based LR system

A score-based LR system needs a scorer and a calibrator. The most basic setup uses a training set and a test set. Both the scorer and the calibrator are fitted on the training set.

import lir
import numpy as np
from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import train_test_split

# generate some data randomly from a normal distribution
X = np.concatenate([np.random.normal(loc=0, size=(100, 1)),
              np.random.normal(loc=1, size=(100, 1))])
y = np.concatenate([np.zeros(100), np.ones(100)])

# split the data into train and test
X_train, X_test, y_train, y_test = train_test_split(X, y)

# initialize a scorer and a calibrator
scorer = LogisticRegression(solver='lbfgs')  # choose any sklearn style classifier
calibrator = lir.KDECalibrator()  # use plain KDE for calibration
calibrated_scorer = lir.CalibratedScorer(scorer, calibrator)

# fit and predict, y_train)
lrs_test = calibrated_scorer.predict_lr(X_test)

# print the quality of the system as log likelihood ratio cost (lower is better)
print('The log likelihood ratio cost is', lir.cllr(lrs_test, y_test), '(lower is better)')
print('The discriminative power is', lir.cllr_min(lrs_test, y_test), '(lower is better)')

# plot calibration
import lir.plotting
lir.plotting.plot_pav(lrs_test, y_test)

The log likelihood ratio cost (CLLR) may be used as a metric of performance. In this case it should yield a value of around .8, but highly variable due to the small number of samples. Increase the sample size to get more stable results.

Project details

Download files

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

Files for lir, version 0.0.8
Filename, size File type Python version Upload date Hashes
Filename, size lir-0.0.8-py3-none-any.whl (28.6 kB) File type Wheel Python version py3 Upload date Hashes View
Filename, size lir-0.0.8.tar.gz (20.3 kB) File type Source Python version None Upload date Hashes View

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring DigiCert DigiCert EV certificate Facebook / Instagram Facebook / Instagram PSF Sponsor Fastly Fastly CDN Google Google Object Storage and Download Analytics Pingdom Pingdom Monitoring Salesforce Salesforce PSF Sponsor Sentry Sentry Error logging StatusPage StatusPage Status page