diagnose issues for labs with unusual heatmap distributions

[mzettersten]

The following labs have heatmaps that appear to have issues:

• babylabAmsterdam
• babylabBrookes
• cecBYU
• childdevlabAshoka
• CorbitLab
• irlConcordia
• kidsdevUniofNewcastle
• PKUSu
• socialcogUmiami

[msschreiner]

It seems to be the case that labs reported the wrong screen dimensions. We are doublechecking if screen dimensions are being tracked within the raw data and editing the information in the new green column screen_dimensions (vs. old red column screen_dimensions_from_labs) of the MB2 Datasets List. Once all labs have been checked and dimensions have been updated, the dataset_sheet in the metadata folder needs to be updated.

[msschreiner]

Could @mzettersten or @alvinwmtan please check the dimensions in the edf files of kidsdevUniofNewcastle with the eyelinkreader package?

[msschreiner]

The screen dimensions of childdevlabAshoka reported by the lab and tracked within their data match. There seems to be another reason for the very unusual heatmap distribution.

[alvinwmtan]

kidsdevUniofNewcastle has 1280x1024 in EDF.

[msschreiner]

Thanks, @alvinwmtan !

[alvinwmtan]

childdevlabAshoka error caused by error in 005 script, fixed here: 330afd8

[alvinwmtan]

Sorry, reopening for a few remaining issues:

• babylabAmsterdam: 5 of the heatmaps cannot render (data issue?)
• childdevlabAshoka: fixed heatmap looks vertically inverted (zero point should be "upper left"?)
• PKUSu: heatmaps look too "zoomed in" (there is almost no looking density at the exits, but there is density around the boxes)
• cclUNIRI: weirdly skewed
• gertlabLancaster: right-shifted

[mzettersten]

@alvinwmtan heatmap for childdevlabAshoka no longer looks inverted to my eye, is that resolved?

[mzettersten]

Double check that we feel comfortable with the (unsurprisingly much messier) web-based heatmaps:

• affcogUTSC
• beinghumanWroclaw
• CogConcordia
• collabUIOWA
• saxelab
• socallabUCSD

Also, the following heatmaps look a bit messier than some others:

• labUNAM

[adriansteffan]

Throwing some more anomalies onto the pile that we could check again:

• SkidLSDLab (e.g. why is IG_LL missing)
• Princeton (e.g. IG_LL and some others are messy)

Will have a look at participant level renders for this

[mzettersten]

cclUNIRI is now fixed!