Weird result difference between cell mode and a straight run

I am writing a program that imputes missing data. So far it contains four cells that pass off to different processes within the imputation. For debugging purposes, I measure the results of the imputation with the original data set using root mean square error (RMSE) values.

When I run the program in cell mode, my RMSE is half the value of running the program straight through. I'm using the same data sets, but getting very different results.

Does anyone know enough about cell mode vs. just calling the program, and why it would behave this way?

Answer 1

[Sean de Wolski]

There should be no difference. Any chance you're calling the same cell twice and it's changing itself?

[Lisa Clayton]

Nope, I've tested it 20 times, 10 using cells and 10 using straight runs. It could be that my RMSE is all over the map in the first place, and the difference is just coincidence, since I pull the occasional large value from the cell mode. But I've yet to pull any small values from the straightforward calls, which made me wonder if there was a difference in the way cell mode handled data vs. just doing a function call.

[Sean de Wolski]

Can you copy and paste the code so we can see it please.

[Sean de Wolski]

There should be no difference