ZeroDivisionError but only sometimes

[luga01]

Hello!
Thank you for making this tool available! I have been using it to obtain m-ratio measures from a Pandas DataFrame consisting of the columns ‘Stimuli’, ‘Responses’, ‘Accuracy’, and ‘Confidence’. The confidence scores of the data are discrete integer scores, ranging from 1-6 (nRatings = 6). (All subjects are treated as a single subject.)

I am able to use the metad() and the .criterion() method without a problem on this DataFrame.

However, when I divide the same dataset up into individual participants and compute the m-ratio for each one, I get a ZeroDivisionError: division by zero, but only after it has successfully computed the m-ratio for several participants.
What I do is create an individual DataFrame for each participant and apply the metad() function to it. This works for a couple of participants and then suddenly turns up with a ZeroDivisionError. I have looked at the DataFrame of the participant for whom I get the error (bottom DataFrame), but cannot spot any difference between it and DataFrames of participants for whom the metad() function does work (top DataFrame).

The first DataFrame works, but for the second one I get the error message:

Do you have any idea why this might be happening or what can be done to fix it?

Thank you in advance,
Lucas Gappmayr

[iezqrom]

Thanks a lot for developing this library. I'm having the same issue and was wondering whether there are any updates to resolve this issue, thanks!

[luga01]

Thanks a lot for developing this library. I'm having the same issue and was wondering whether there are any updates to resolve this issue, thanks!

Hey. You can try the hmetad() function instead. That one might work. I remember managing to run this function in a conda environment. My code looked like this:

result = hmetad(data = df, nRatings = 6, stimuli = 'Stimuli', accuracy = 'Accuracy',
confidence = 'Confidence', output = 'dataframe', padAmount = 1, padding = True, verbose = 0)

[iezqrom]

Thanks! I ended up using the code from the 'competition' in Matlab (hMeta-d toolbox)... hehe

[LegrandNico]

@luga01 could you send a data frame that reproduces the error?

[lenarddome]

I am just jumping on the bandwagon here. Here is some data to reproduce the error:
buggy.csv

results = metadpy.mle.metad(
    data=data,
    nRatings=4,
    stimuli="signal",
    accuracy="accuracy",
    confidence="bins",
    verbose=0,
)