Imputation Missing Value with Regression

Hai, i'm new in here. I'm doing my final project now. Actually i'm not in statistics department or something about statistics. I'm in informatics faculty.
It's about imputation missing value. And the method i used here is about regression. But actually i can't really get it how the method works in my dataset.
From the journal that i've read, it's about get the value of variance, SSE, and predict parameter of B.
Below, i attached the file about how the method works. But i still can't get it when i have to use it in my dataset.
Anybody can help me? :(
Thanks before
i'm using predictive mean matching method.
from what i've read, this method apply the linear regression *just like what the paper explain* for the process.
but i still can't understand how this process works in my dataset.

FYI: I'm using Matlab to built a such kind of application that can impute the missing values from the dataset. Then, i will analyse how the method works with classfication concept of Data Mining.

And so, i'm stuck with how this linear regression works in my dataset :(

I hope there'll be somebody help me about this :(
Thanks before
then i'm forget to share about the previous page of that paper's page above. so it's here which contain some of information of the symbols.


Ambassador to the humans
That wasn't the part that got me - I've never done PMM but it appears that the first thing you posted doesn't fully describe the method and there is more to it. Is that so?
Hmm in the paper, the part that explain how the method work is only to that part. because it's only concern about compared some of methods.
but if you mind to read it, i will send to your email. the file is little too big to be attached in here.

thanks before :')


Ambassador to the humans
Well I just don't think I could help without know more about how the set of possible donors is created.
same with me, i can't get that too.
anyway, do you know about the dataset that i can use to implement this regression method?
is there any criteria of the dataset? or maybe would you mind to share it with me?