Multiple Regression for imputation missing value which described by NaN

Is there somebody know about imputation missing value with regression or even multiple regression method? For example, i have this matrix with missing value which described by NaN:

     2     1     4     6     2
     9     4     6     1     2
     5     3     2     8     3
     7     2     1   NaN     3
     7   NaN   NaN     2     4

How can i use the regression method to replace the NaN? Please help me understand about that with the matlab code. Thanks before :')

Answer 1

[Walter Roberson]

Consider using the MATLAB File Exchange contribution inpaint_nans

[Isti]

he said that he doesn't use regression method for that.

so i still can't find the solution :(

anyway, thanks for sharing.