OHDSI Home | Forums | Wiki | Github

Prediction problem in atlas: T size and O count all fall to 0 when choosing neural network in analysis setting


(panpan) #1

dear all:
when I choose the lasso logistic model to predict in analysis setting, the size and count of the target cohort and outcome cohort are separately 5000 and 300, and the shiny will give me the model table. but when I choose the other model such as Multilayer Perceptron, or k-Nearest Neighbor, the T size and O count all fall to 0, it is very hard to understand, right?
what is the reason?

thank you very much

(Chungsoo Kim) #2

Hi @pandamiao

I think If you see 0 patients in shiny viewer,
it may mean that the model development has not proceeded, not a cohort generation problem.
Please check the log file or model file in outputFolder.
If you cannot find model file in outputFolder, then model development was not in proceeded.
Also check the package installation(eg. KNN needs BigKnn package) or version.

Chungsoo Kim

(panpan) #3

thank you very much @Chungsoo_Kim.