Taher Babaee , in SPSS, one option will be:

0) Assume you have missing data on a variable of interest Y, but in turn you have data in X for the missing observations and the non-missing observations of Y.

1) Then estimate in SPSS (v27) a Bayesian regression of Y_nm against X_nm, where X_nm is the subset of observations for which there is in non-missing (nm) information in Y:

Y_nm = a + b*X_nm

2) Then you can use the estimates of a,b in the regression in (1) to estimate the missing values Y_m:

Y_m = p(a) + p(b)*X_m

Note that as this is a Bayesian estimation, p(a) and p(b) will not be point estimates, but rather full p.d.fs. Point estimates can be obtained assuming e.g. quadratic lost.

See:

https://www.ibm.com/docs/en/spss-statistics/27.0.0?topic=models-bayesian-linear-regression-predict

And some applications of mine here:

Article Modeling Hyperinflation Phenomenon: A Bayesian Approach

See these links:

https://www.researchgate.net/topic/Missing-Data/3

https://www.youtube.com/watch?v=ytQedMywOjQ

Dear @Remal Al-Gounmeein and @Rolando Gonzales

Many thanks for your response.