Statistics Toolbox

Functions for Data with Missing Values (NaNs)

Most real-world data sets have one or more missing elements. It is convenient to code missing entries in a matrix as NaN (Not a Number).

Here is a simple example.

• m = magic(3);
  m([1 5]) = [NaN NaN]
  
  m =
      NaN      1      6
        3    NaN      7
        4      9      2
  

Any arithmetic operation that involves the missing values in this matrix yields NaN, as below.

• sum(m)
  
  ans =
      NaN   NaN     15
  

Removing cells with NaN would destroy the matrix structure. Removing whole rows that contain NaN would discard real data. Instead, the Statistics Toolbox has a variety of functions that are similar to other MATLAB functions, but that treat NaN values as missing and therefore ignore them in the calculations.

• nansum(m)
  
  ans =
       7    10    13
  
  

  NaN Functions
  nanmax	Maximum ignoring NaNs
  nanmean	Mean ignoring NaNs
  nanmedian	Median ignoring NaNs
  nanmin	Minimum ignoring NaNs
  nanstd	Standard deviation ignoring NaNs
  nansum	Sum ignoring NaNs

In addition, other Statistics Toolbox functions operate only on the numeric values, ignoring NaNs. These include iqr, kurtosis, mad, prctile, range, skewness, and trimmean.

Measures of Dispersion

Function for Grouped Data