ecdf

Compute empirical cumulative distribution function (ecdf) for baseline and target data specified for drift detection

Syntax

E = ecdf(DDiagnostics)

E = ecdf(DDiagnostics,Variables=variables)

Description

E = ecdf(DDiagnostics) returns the table E, which stores the ecdf values for all the variables specified for drift detection in the call to detectdrift.

ecdf returns NaN values for categorical variables.

E = ecdf(DDiagnostics,Variables=variables) returns the table E for the variables specified by variables.

Examples

Compute Empirical Cumulative Distribution Function (ECDF) for All Variables

Generate baseline and target data with two variables, where the distribution parameters of the second variable change for target data.

rng('default') % For reproducibility
baseline = [normrnd(0,1,100,1),wblrnd(1.1,1,100,1)];
target = [normrnd(0,1,100,1),wblrnd(1.2,2,100,1)];

Perform permutation testing for any drift between the baseline and target data.

DDiagnostics = detectdrift(baseline,target)

DDiagnostics = 
  DriftDiagnostics

              VariableNames: ["x1"    "x2"]
       CategoricalVariables: []
                DriftStatus: ["Stable"    "Drift"]
                    PValues: [0.2850 0.0030]
        ConfidenceIntervals: [2x2 double]
    MultipleTestDriftStatus: "Drift"
             DriftThreshold: 0.0500
           WarningThreshold: 0.1000


  Properties, Methods

Compute the ecdf values for both variables.

E = ecdf(DDiagnostics)

E=2×3 table
                x             F_Baseline         F_Target   
          ______________    ______________    ______________

    x1    {201x1 double}    {201x1 double}    {201x1 double}
    x2    {201x1 double}    {201x1 double}    {201x1 double}

E is a table with two rows and three columns. The two rows correspond to the two variables. For each variable, ecdf computes the ecdf values over a common domain for baseline and target data. It stores the common domain for each variable in the column x, the ecdf values for baseline data in the column F_Baseline, and the ecdf values for target data in the column F_Target.

Access the ecdf values for variable 2 in baseline data.

E.F_Baseline{2}

ans = 201×1

         0
    0.0100
    0.0100
    0.0200
    0.0300
    0.0400
    0.0500
    0.0600
    0.0700
    0.0800
      ⋮

Plot the empirical cumulative distribution function values of baseline and target data for variable x2.

stairs(E.x{2},E.F_Baseline{2},LineWidth=1.5)
hold on
stairs(E.x{2},E.F_Target{2},LineWidth=1.5)
title('ECDF for x2')
xlabel('x2')
ylabel('Empirical CDF')
legend('Baseline','Target',Location = 'east')
hold off

The plot of the ecdf values also show the drift in the distribution of the target data.

Copyright 2021 The MathWorks, Inc.

Compute ECDF Values for Specified Variables

Load the sample data.

load humanactivity

For details on the data set, enter Description at the command line.

Assign the first 1000 observations as baseline data and next 1000 as target data.

baseline = feat(1:1000,:);
target = feat(1001:2000,:);

Test for drift on all variables.

DDiagnostics = detectdrift(baseline,target);

Compute the ecdf values for only the first five variables.

E = ecdf(DDiagnostics,Variables=[1:5])

E=5×3 table
                 x             F_Baseline          F_Target    
          _______________    _______________    _______________

    x1    {2001x1 double}    {2001x1 double}    {2001x1 double}
    x2    {2001x1 double}    {2001x1 double}    {2001x1 double}
    x3    {2001x1 double}    {2001x1 double}    {2001x1 double}
    x4    {2001x1 double}    {2001x1 double}    {2001x1 double}
    x5    {2001x1 double}    {2001x1 double}    {2001x1 double}

Access the ecdf values for the third variable in baseline data.

E.F_Baseline{3}

ans = 2001×1

         0
         0
         0
         0
         0
         0
    0.0010
    0.0020
    0.0030
    0.0040
      ⋮

Plot the empirical cumulative distribution function values of baseline and target data for variable x3.

stairs(E.x{3},E.F_Baseline{3},LineWidth=1.5)
hold on
stairs(E.x{3},E.F_Target{3},LineWidth=1.5)
title('ECDF for x3')
xlabel('x3')
ylabel('Empirical CDF')
legend('Baseline','Target',Location = 'southeast')
hold off

The ecdf plot shows the drift in the target data for variable x3.

Input Arguments

DDiagnostics — Diagnostics of the permutation testing for drift detection

DriftDiagnostics object

Diagnostics of the permutation testing for drift detection, specified as a DriftDiagnostics object returned by detectdrift.

variables — List of variables

string array | cell array of character vectors | integer indices

List of variables for which to compute the ecdf values, specified as a string array, a cell array of character vectors, or a list of integer indices.

Example: Variables=["x1","x3"]

Example: Variables=(1,3)

Data Types: single | double | char | string

Output Arguments

E — ecdf values

table

ecdf values for all variables specified for drift detection in the call to detectdrift, returned as a table with the following columns.

Column name	Description
x	Common domain over which to evaluate the empirical cdf
F_Baseline	ecdf values for the baseline data
F_Target	ecdf for the target data

For each variable in E, the columns hold x and the ecdf values in cell arrays. To access the values, you can index into the table; for example, to obtain the ecdf values for the second variable in baseline data, use E.F_Baseline{2,1}.

Version History

Introduced in R2022a