Increasing memory, exact logistic regression

[John Gustavsson]

Hi,

I'm using STATA 16. I'm trying to do an exact logistic regression, but it gives me the error message "exceeded memory limit of 10.0M bytes; use the memory() option to increase the
memory limit"

However, I've read the instructions and I don't get it... say I want to increase the memory to for example 30 megabyte, what exactly would I type into the command box?

I'm grateful for any help.

Merry Christmas!

[William Lisowski]

The output of help exlogistic for its memory() option tells us that the default is "memory(10m), where m stands for megabyte".

You ask "what exactly would I type into the command box" to increase the memory to 30 megabytes, which makes me think perhaps you are running exlogistic from the Stata menus by selecting Statistics > Binary outcomes > Exact logistic regression. When I choose that, the dialog box that opens has a row of tabs. I click on the Options tab and the following appears in the window.


In the box that contains "10" next to the label "Limit the memory usage in megabytes" replace the 10 with 30.

On the other hand, if you are typing the exlogistic command into Stata's Command window, you would add the option memory(30m) to the end of your command. I can't be more helpful than that, because I don't know what you are already typing into Stata's command window. Do understand that if you have no other options, you need to add a comma to the end of your current command and then follow it with memory(30m) to tell Stata that what follows are the options.

[Joseph Coveney]

I want to increase the memory to for example 30 megabyte, what exactly would I type into the command box?

Maybe try

 exlogistic . . ., memory(30m)

[John Gustavsson]

Thanks so much

That did work, the problem was once I added some control variables even 2 gigabyte of memory was not enough. Is there any way to "get around" this issue or will I just have to do something other than exact logistic regression?

[Lara Nuessle]

I would be also interested in this question John posted at the end. I have a sample size of 900 observations, however with 25 variables (true controlls, but also many industry dummies).

I regressed already with the "normal" logit function but the results are pretty overall insignificant which i cannot explain myself hence i relied on logical controlls whihc other papers also used. When trying exlogistics however i also reached the 2 GB memeory without reaching to a result.