I need tp find the missing values y1 and y2
- I have 5 sources of seed (1,2,3,4,5) and four grounds(I,II,III,IIV)
| I | II | III | IV |
1 | 157 | 162 | 154 | 165 |
2 | 163 | 173 | 169 | 171 |
3 | y1 | 161 | 170 | 167 |
4 | 178 | 173 | 180 | 182 |
5 | 165 | 175 | y2 | 181 |
<tbody>
</tbody>
So can you help to know how by using excel or any other programe??
thanx
Nourali,
The problem you pose is very reasonable. It's a fairly standard statistical problem. And yes, you do give enough information to get the results you request.
Set up the model
y(i,j)= S(i) + G(j) + random error.
You give 18 outcome values for the y(i,j) (because 2 are unknown) and have 9 parameters to determine, viz. the "effects" of the 5 seed sources and of the 4 types of ground.
This model has a loss of one degree of freedom because the parameters have one linear constraint. This is generally easily resolved by putting one of them equal to zero.
So available degrees of freedom in the model are 18 - 9 - 1.
This is >0 so the effects can be estimated and your missing values predicted by several statistical techniques.
Sometimes this type of model is extended by including an interactive effect, whereby the model becomes
y(i,j)= S(i) + G(j) + S(i)*G(j) + random error
somewhat more complex and usually not done without good reason for supposing the interactive effect may be real.
I've done this sort of thing in the past by variants of the standard regression (Excel's LINEST etc) and if you get your data layout right it's actually quite easy.
But I won't do any more without knowing just why you want to do this, what is your statistical background, and what other sources have you investigated for a possible solution?