how to run multiple number of regressions?

I have table T with

Date Location Sales Visitors
1/1 NY 10 5
1/1 LA 7 3
1/2 NY 9 5
1/2 LA 4 1

Then I want to run regressions

(sales)=a_i*(visitor)+b_i

for each i = {NY, LA}. Is there any simpler way to do this?

Answer 1

[dpb]

tRegress=rowfun(@(x,y)polyfit(x,y,1),T,'GroupingVariables',{'Location'}, ...
                                       'InputVariables',{'Visitors','Sales'}, ...
                                       'NumOutputs',1, ...
                                       'OutputVariableNames',{'Coefficients'});

[alpedhuez]

can one run on fitlm? It is because the actual data is more complex than the example above.

[dpb]

tRegress=rowfun(@(x,y)fitlm(x,y,'linear'),T,'GroupingVariables',{'Location'}, ...
                                           'InputVariables',{'Visitors','Sales'},  ...
                                           'NumOutputs',1, ...
                                           'OutputVariableNames',{'Model'}, ...
                                           'OutputFormat',"cell");

If it gets too much more complicated may have to use an m-file for the function instead of anonymous function...also have to switch to a cell output to catenate the model outputs instead of table.

[alpedhuez]

Let me work on it.