topic Weight of specific group of data in whole dataset in Discussions

Weight of specific group of data in whole dataset

Gabriel — Sat, 10 Jun 2023 23:40:48 GMT

In the image attached, I am trying to find out or rather calculate weight of the set of variables in circle in (A). When it is included in the data set it gives an R-sq of 0.59, however, when it is excluded the coefficient reduces to the minimum (B).

How can give a weighted value to the dataset, as I would like to exclude it from the dataset while showing and explaining on valid reasons why it is excluded?

Re: Weight of specific group of data in whole dataset

txnelson — Thu, 18 Nov 2021 21:49:21 GMT

If I correctly understand what you want to do, all you have to do, is to go to the graph and drag and select the data values in question, and then right click on one of the selected points and in Graph Builder or Fit Y by X

Rows=>Hide and Exclude

Re: Weight of specific group of data in whole dataset

jthi — Fri, 19 Nov 2021 09:13:08 GMT

Other guess depending what you mean with the weight / exclusion:

Using big class as an example:

Red points are the ones I want to exclude from analysis (but still how them):

R2 = 0.559

Select them and just exclude (don't hide):

They will disappear from graph builder:

R2:0.351

Now HOLD SHIFT while clicking red triangle and it will show you hidden options (why are these hidden and are there more hidden options / list of hidden options some where?) (found from here). The excluded rows should be shown a bit transparent

R2:0.351

Re: Weight of specific group of data in whole dataset

Gabriel — Fri, 19 Nov 2021 16:46:34 GMT

Thank you, your resonspe helped in showing the group of excluded data. However, one of the lines of fit suddenly is not showing even when I redo the graph.

Also, the question was partly answered too, How can I do a statistical calculation to see how much this three group influenced the fit (which is an expected outcome), hence, one of the reasons I have excluded them?

Any idea how I can present this?