IALSA / IALSA-2015-Portland

Hub to accompany IALSA 2015 workshop at Portland, OR, Feb 22-25, 2015
GNU General Public License v2.0
2 stars 0 forks source link

Matching wide to long #135

Closed andkov closed 8 years ago

andkov commented 8 years ago

@wibeasley , some of values don't match between long and wide formats. Here's an example. Any idea what might be the issue?

image

wibeasley commented 8 years ago

Thanks for checking on me. There's a lot of moving parts in 110k rows, it gets confusing so I easily could have made a mistake. However in this case, I think everything is working correctly. That first/long table is already filtering on 'gait' (otherwise the 110k rows were freezing the browser and it was unusable).

Because a filter is already in place, your first screen shot is showing only gait x block, and not showing all three of 'gait x block' | 'grip x block' | 'perf+block'. To get the window you want, you have to step inside the program.

image image

image

I'm glad you brought this up, because this shows the scenario that supports @ampiccinin & @GracielaMuniz's desire to collapse the block growth model parameters across multiple models (ie, paired with gait, grip, & pef).

wibeasley commented 8 years ago

If you want to compare yourself, you have to be content with something crossed with gait, such as:

est

image

se

image