Fix issue 338 - pspm_glm

[teddychao]

Fixes #338.

Changes proposed in this pull request:

• Add a warning when input data was filtered out more than 10% due to missing epoch detection.
• Specify the reason of missing epochs in the warning after checking sampling rates.

[dominikbach]

I'd say "is possibly caused by downsampling" - the data might just as well have been missing in the first place, before downsampling.

[m-guggenmos]

Just to make sure, is this line in the fix correct?

perc_missing = 1 - sum(glm.M)/length(glm.M);

I'm getting a warning related to this variable a lot and intuitively it makes more sense without the 1 -? I'm not familiar with the codebase though.

[teddychao]

Just to make sure, is this line in the fix correct?

perc_missing = 1 - sum(glm.M)/length(glm.M);

I'm getting a warning related to this variable a lot and intuitively it makes more sense without the 1 -? I'm not familiar with the codebase though.

Hi @m-guggenmos,
Thank you for propose the issue to us. Yes it is a mistake. The positive values in M and glm.M denote missing values, thus the perc.missing should not contain a 1- there.
We apologise for any inconenience the mistake brought to you. We will include this fix in our next release that will be available very soon.
Best wishes
Teddy

[m-guggenmos]

Great, thanks for the prompt response!