peak-engines 0.2.0

Build Warped Linear Regression Models

peak-engines is a Python package for building Warped Linear Regression models.

Warped Linear Regression is like Ordinary Least Squares but with an extra transformation step where target values are remapped using a parameterized monotonic function and adjusted so as to maximize likelihood on a linear model. The transformation makes Warped Linear Regression more general purpose than Ordinary Least Squares and able to fit models with non-normal error distributions.

For more details on the math behind Warped Linear Regression models see What to Do When Your Model Has a Non-Normal Error Distribution.

Installation

pip install peak-engines

Getting started

Load an example dataset

from sklearn.datasets import load_boston
X, y = load_boston(return_X_y=True)

Split out training and testing portions

from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(
  X, y, test_size=0.1, random_state=0)

Fit a Warped Linear Regression Model

import peak_engines
model = peak_engines.WarpedLinearRegressionModel()
model.fit(X_train, y_train)

Visualize the warping function

import numpy as np
import matplotlib.pyplot as plt
y_range = np.arange(np.min(y), np.max(y), 0.01)
z = model.warper_.compute_latent(y_range)
plt.plot(y_range, z)
plt.xlabel('Median Housing Value in $1000s')
plt.ylabel('Latent Variable')
plt.scatter(y, model.warper_.compute_latent(y))

Make predictions

from sklearn.metrics import mean_squared_error
y_pred = model.predict(X_test)
print(mean_squared_error(y_test, y_pred))
41.56037297819257

Compute and plot the error distribution of a prediction

logpdf = model.predict_logpdf([X_test[0]])
def pdf(yi):
    return np.exp(logpdf([yi]))
plt.plot(y_range, [pdf(yi) for yi in y_range])
plt.axvline(y_test[0])
plt.xlabel('Median Housing Value in $1000s')
plt.ylabel('Probability Density')

Examples

• example/boston_housing.ipynb: Build a model to predict housing values.

• example/fish.ipynb: Predict the weight of fish.

• example/abalone.ipynb: Predict the age of sea snails.

Documentation

See doc/Reference.pdf.