How to get Max Values for Data Subsets

Question

Hi community,

I'm running into issues when I want to continue with my pearson correlation output. I use the pearson correlation as initial step for my data analysis to investigate the relation of all paramters to my target variable. The data is structured in subsets of classes as I want to find out the best configuration of each class in terms of having the highest impact on the target variable.This is where I get problems: I want to identify the highest correlation based on the class, but I can't figure out how to solve this. I tried it with filtering, and summarizing, but can't achieve to get what I want.

I provide some test data. As you can see, it is structured for each class in the same way, e.g. sun, and then I have some variations for sun and all other classes. Somehow, negative correlation should also be considered, thinking of max of absolut values.

The desired result, which I want to continue working with would look like this:

FieldNameMaxCorrValuesun_1_0000,213124rain_0_5000,214803hail_1_0000,054369fog_1_0000,200903

Thanks for any help with this issue.

Theresa

pearson_correlation_output.xls

Felipe_Ribeir0 · Answer

Hi @Thess_Luck

The join was made just to compare the result until the join with the expected result that you provided!

You can see the result for each column after the select:

How to get Max Values for Data Subsets.yxmd

Thess_Luck · Answer

Hey, it seems to me as in your proposed workflow you need the right join input already defined. The sample of the expected result needs to be known in advance, but what if that is not the case, because going throug the table manually is exactly what I want to get around.

Felipe_Ribeir0 · Answer

Hi @Thess_Luck

Please see if this is correct:

How to get Max Values for Data Subsets.yxmd