Mottl / longtail

Longtail transforms RV from the given empirical distribution to the standard normal distribution

longtail

Longtail transforms RV from the given empirical distribution to the standard normal distribution.

Installation

pip install -e git+https://github.com/Mottl/longtail.git#egg=longtail

Usage

import numpy as np
import longtail

x = np.random.laplace(size=100000)
longtail.plot(x)

Estimating distributions parameters...
norm (0.0018876631529621596, 1.4211949719613757)
laplace (0.0017959270327248976, 1.004473511770026)
cauchy (0.0008371221028761951, 0.6495115549176855)

scaler = longtail.GaussianScaler()
x_ = scaler.fit_transform(x)
longtail.plot(x_)

Estimating distributions parameters...
norm (5.4850160001935534e-05, 0.9999632908186453)
laplace (5.727594101276392e-05, 0.7978789360688088)
cauchy (-1.422402203800512e-06, 0.6119807936005598)

plt.plot(scaler.transform_table[:,0], scaler.transform_table[:,1],
    color="dodgerblue", label="Laplace to Gaussian transformation")
plt.title("Transformation function")
plt.xlabel(r"$x$")
plt.ylabel(r"$\hat{x}$", rotation=0)
plt.axis("equal")
plt.grid(True)
plt.show()

Requirements

• Python 3
• numpy
• scipy
• matplotlib