Using Linear Regression - running all independent variables.

[HenryE]

Hi,

I am trying to investigate the relationship between an independent variable and about 90 dependent variables using linear regresssion. What  I have done in the past is to run one independent ( or a few) against the dependent variable, determine the relationship, then keep adding independent variables one after another and running it till I get to the final independent variables having a significant effect on the dependent variable. 

The problem with this is that, this is very time consuming. I was made aware that there is a way to run all the independent variables at once and Alteryx will find the optimal significant independent variables. 

Could somebody let me know how to go about this. 

Thanks,

[OldDogNewTricks]

I believe what you are looking for is the 'Association Analysis Tool' (link). Within this tool there is a selection that states 'Target a field for more detailed analysis'.  Once this is selected, you can choose a target field which creates correlation between the different variables as well as an interactive correlation plot.

Another possibility would be the 'Spearman correlation Tool' (link).

I have attached a workflow with both tools in use and applied to the UC Irvine mpg data set.

If you are looking for determining 'real importance' within a linear model by running multiple iterations of the model with variables in different order, then you'll have to write your own implementation in R (which I have an example of as well).  There is a library that already exists that does the heaving lifting for that as well (link).

[HenryE]

Thanks. I will give it a go

[OldDogNewTricks]

Ok.  Well let me know.

[OldDogNewTricks]

Just curious if this answered your question?

[HenryE]

Hello,

Yes it was very helpful. I used the association analysis tool. I went ahead to use the stepwise tool.

Thanks,

[OldDogNewTricks · Accepted solution]

Would you mind marking my answer as the solution and the problem solved?