Read the Medium Blog for more information
distfit is a python package for probability density fitting of univariate distributions for random variables.
With the random variable as an input, distfit can find the best fit for parametric, non-parametric, and discrete distributions.
-
For the parametric approach, the distfit library can determine the best fit across 89 theoretical distributions. To score the fit, one of the scoring statistics for the good-of-fitness test can be used used, such as RSS/SSE, Wasserstein, Kolmogorov-Smirnov (KS), or Energy. After finding the best-fitted theoretical distribution, the loc, scale, and arg parameters are returned, such as mean and standard deviation for normal distribution.
-
For the non-parametric approach, the distfit library contains two methods, the quantile and percentile method. Both methods assume that the data does not follow a specific probability distribution. In the case of the quantile method, the quantiles of the data are modeled whereas for the percentile method, the percentiles are modeled.
-
In case the dataset contains discrete values, the distift library contains the option for discrete fitting. The best fit is then derived using the binomial distribution.
Documentation pages
On the documentation pages you can find detailed information about the distfit library with many examples.
Installation
Install distfit from PyPI
pip install distfitInstall from github source (beta version)
install git+https://github.com/erdogant/distfitCheck version
import distfit
print(distfit.__version__)The following functions are available after installation:
# Import library
from distfit import distfit
dfit = distfit() # Initialize
dfit.fit_transform(X) # Fit distributions on empirical data X
dfit.predict(y) # Predict the probability of the resonse variables
dfit.plot() # Plot the best fitted distribution (y is included if prediction is made)Examples
Example: Quick start to find best fit for your input data
# [distfit] >INFO> fit
# [distfit] >INFO> transform
# [distfit] >INFO> [norm ] [0.00 sec] [RSS: 0.00108326] [loc=-0.048 scale=1.997]
# [distfit] >INFO> [expon ] [0.00 sec] [RSS: 0.404237] [loc=-6.897 scale=6.849]
# [distfit] >INFO> [pareto ] [0.00 sec] [RSS: 0.404237] [loc=-536870918.897 scale=536870912.000]
# [distfit] >INFO> [dweibull ] [0.06 sec] [RSS: 0.0115552] [loc=-0.031 scale=1.722]
# [distfit] >INFO> [t ] [0.59 sec] [RSS: 0.00108349] [loc=-0.048 scale=1.997]
# [distfit] >INFO> [genextreme] [0.17 sec] [RSS: 0.00300806] [loc=-0.806 scale=1.979]
# [distfit] >INFO> [gamma ] [0.05 sec] [RSS: 0.00108459] [loc=-1862.903 scale=0.002]
# [distfit] >INFO> [lognorm ] [0.32 sec] [RSS: 0.00121597] [loc=-110.597 scale=110.530]
# [distfit] >INFO> [beta ] [0.10 sec] [RSS: 0.00105629] [loc=-16.364 scale=32.869]
# [distfit] >INFO> [uniform ] [0.00 sec] [RSS: 0.287339] [loc=-6.897 scale=14.437]
# [distfit] >INFO> [loggamma ] [0.12 sec] [RSS: 0.00109042] [loc=-370.746 scale=55.722]
# [distfit] >INFO> Compute confidence intervals [parametric]
# [distfit] >INFO> Compute significance for 9 samples.
# [distfit] >INFO> Multiple test correction method applied: [fdr_bh].
# [distfit] >INFO> Create PDF plot for the parametric method.
# [distfit] >INFO> Mark 5 significant regions
# [distfit] >INFO> Estimated distribution: beta [loc:-16.364265, scale:32.868811]
Example: Plot summary of the tested distributions
After we have a fitted model, we can make some predictions using the theoretical distributions. After making some predictions, we can plot again but now the predictions are automatically included.
Example: Make predictions using the fitted distribution
Example: Test for one specific distributions
The full list of distributions is listed here: https://erdogant.github.io/distfit/pages/html/Parametric.html
Example: Test for multiple distributions
The full list of distributions is listed here: https://erdogant.github.io/distfit/pages/html/Parametric.html
Example: Fit discrete distribution
from scipy.stats import binom
# Generate random numbers
# Set parameters for the test-case
n = 8
p = 0.5
# Generate 10000 samples of the distribution of (n, p)
X = binom(n, p).rvs(10000)
print(X)
# [5 1 4 5 5 6 2 4 6 5 4 4 4 7 3 4 4 2 3 3 4 4 5 1 3 2 7 4 5 2 3 4 3 3 2 3 5
# 4 6 7 6 2 4 3 3 5 3 5 3 4 4 4 7 5 4 5 3 4 3 3 4 3 3 6 3 3 5 4 4 2 3 2 5 7
# 5 4 8 3 4 3 5 4 3 5 5 2 5 6 7 4 5 5 5 4 4 3 4 5 6 2...]
# Import distfit
from distfit import distfit
# Initialize for discrete distribution fitting
dfit = distfit(method='discrete')
# Run distfit to and determine whether we can find the parameters from the data.
dfit.fit_transform(X)
# [distfit] >fit..
# [distfit] >transform..
# [distfit] >Fit using binomial distribution..
# [distfit] >[binomial] [SSE: 7.79] [n: 8] [p: 0.499959] [chi^2: 1.11]
# [distfit] >Compute confidence interval [discrete]
Example: Make predictions on unseen data for discrete distribution
Example: Generate samples based on the fitted distribution
Contributors
Setting up and maintaining distfit has been possible thanks to users and contributors. Thanks:
Citation
Please cite distfit in your publications if this is useful for your research. See column right for citation information.