Looking back there are definitely a few things I would have done differently. First of all, the data being sent over — as of now I send over all the values of the built in accelerometer (heading, pitch, and roll) when the only value I am using is roll, where I see if it’s between 85 to 95 (which means yes thesis) or -85 to -95 (which means no thesis). And all values in between are in inclusive states (cube was probably being handled).
Another thing I would have changed would be to use a wall power adapter rather than a 9V, so I didn’t have to worry about the cube losing power.
Lastly, I’d use a different sensor set up where it didn’t rely on me remembering to adjust its state.
Here I graphed the data, using the date column and the roll values.
It may look like I spent an astronomical amount of time on my thesis, but don’t be impressed, I was sending data every 3 seconds. There is also gaps between days where there was no data being sent … also meaning, I wasn’t working on my thesis.