Muse data is frequently -1000 for TP9 and TP10

[ErikBjare]

Not sure what's up with that, or how to deal with it.

From looking at the raw data, it looks like it's -1000 exactly every 5th row. Sometimes there are 2 in a row, and then it repeats every 5th row again.

Edit: Maybe this is just powerline noise? At the sampling freq of 250Hz the powerline peak would happen roughly every 4-5th sample. Why are TP9 and TP10 so much more sensitive though?

Example:

1603711387.314,-1000.000,-44.434,-38.574,-1000.000,0.000
1603711387.318,-609.375,-29.297,-27.344,-574.707,0.000
1603711387.322,787.109,-19.531,-23.926,814.941,0.000
1603711387.325,-852.051,-27.832,-22.461,-858.887,0.000
1603711387.329,184.082,-37.598,-23.926,189.941,0.000
1603711387.333,-1000.000,-45.410,-39.062,-1000.000,0.000
1603711387.337,-836.914,-34.668,-30.762,-804.688,0.000
1603711387.341,519.043,-18.555,-11.719,561.523,0.000
1603711387.345,-801.758,-18.555,-20.508,-808.105,0.000
1603711387.349,150.391,-23.438,-27.832,155.762,0.000
1603711387.353,-1000.000,-29.785,-26.367,-1000.000,0.000
1603711387.357,-1000.000,-30.762,-27.832,-1000.000,0.000
1603711387.361,178.711,-21.484,-20.508,231.934,0.000
1603711387.365,-764.648,-22.949,-21.484,-768.555,0.000
1603711387.368,222.168,-33.203,-31.250,198.242,0.000
1603711387.372,-1000.000,-39.551,-32.715,-1000.000,0.000
1603711387.376,-1000.000,-33.203,-31.250,-1000.000,0.000
1603711387.380,-36.133,-21.484,-26.367,-6.836,0.000
1603711387.384,-789.062,-16.113,-27.832,-781.738,0.000
1603711387.388,409.668,-18.066,-33.691,378.906,0.000
1603711387.392,-909.180,-28.320,-34.180,-925.293,0.000
1603711387.396,-1000.000,-26.855,-33.203,-1000.000,0.000
1603711387.400,-213.867,-16.113,-29.785,-186.523,0.000
1603711387.404,-873.535,-15.137,-23.438,-854.492,0.000
1603711387.407,650.879,-21.484,-28.320,603.516,0.000
1603711387.411,-738.281,-66.406,-38.086,-779.785,0.000
1603711387.415,-1000.000,-78.613,-34.180,-1000.000,0.000
1603711387.419,-204.102,-25.879,-19.531,-210.449,0.000
1603711387.423,-943.848,-7.812,-18.555,-930.664,0.000
1603711387.427,878.418,-9.277,-25.879,835.938,0.000
1603711387.431,-439.453,-28.320,-37.109,-500.977,0.000
1603711387.435,-1000.000,-36.621,-27.344,-1000.000,0.000
1603711387.439,-177.734,-18.066,-20.996,-181.641,0.000
1603711387.443,-991.699,-17.578,-37.598,-982.910,0.000
1603711387.447,-974.121,-29.297,-38.574,-996.094,0.000
1603711387.450,-139.160,-31.738,-42.969,-194.824,0.000
1603711387.454,-1000.000,-18.066,-48.340,-1000.000,0.000
1603711387.458,-303.223,-12.695,-33.691,-268.066,0.000

[ErikBjare]

It appears this problem does not appear in data/eeg/muse/subject0000/session001/recording_2020-09-30-09.18.34.csv.

It could have something to do with simultaneous PPG collection (#9), or something else entirely (like simply a bad fit/conductance). In any case, I've disabled simultaneous PPG collection in 9f14a58.

[ErikBjare]

This seems to be purely a case of bad signal. I've added a preprocessing/cleaning step in cf6b56b that removes rows that have unusually high/low values like this.