Day 360: Second experiment

The computer I use to process all my data, even with this it takes FOREVER to finish processing.

Well tomorrow is my second experiment. To be clear, it’s technically the same experiment, just a second person, so maybe it would be better to call it trial two? In any case, while the last one went (mostly) well, there are a few things I’m changing to make this next one even better. Sometimes when you prep for an experiment, you don’t realize what you’re forgetting until you can’t do it, so we’re correcting that this time.

When we record our data, we are using two different sets of sensors that record at their own frequencies. This is a problem when we analyze the data because if in one second we have 1000 samples for one set of sensors and 500 samples for one second for a second set of sensors, we need to either upsample the 500 samples to match the 1000 samples, or downsample the 1000 samples to match the 500 samples. I prefer to upsample, but that’s just me.

Why am I telling you all this? Well we need to align our data across several different recordings. Since we start them all separately there is no way to make sure they are aligned right out of the gate. To solve this problem we throw in flags across our recordings to align everything. A flag is just a little note that we trigger by some action, pressing a button, using a stimulator, etc. These flags happen at the exact same time across our sensors so we have a way to align the data when the time comes to analyse it.

Additionally we throw in start and stop flags to let us know when we’re doing the experiment and when we are doing something else. We don’t stop the recording because that would mean we need to realign everything as we go and that’s never a good thing. It introduces another source of error and we want to minimize that.

Long story short, when I created the workspace to do the recordings, I also set up several predefined flags I could throw into the data. Being the super intelligent person that I am, I created one flag for each of our conditions and an oops flag (literally it would annotate the data with an oops). Notice what I forgot yet? I forgot a stop flag. So I used the condition flag twice once at the start and once at the end.

This is going to make segmenting my data a huge pain since I need to do this blindly (IE matlab). We’re sampling at 1000 samples a second so I have ~3,600,000 samples an hour of recording, with about an hour and a half of recording per person. Luckily it’s just subject 1, so I will fix this problem and the next four subjects SHOULD be easier to deal with.

There were some other oops along the way. For example the way I attached the sensors could be improved upon, the ground and reference electrodes were placed in a bad spot for one of the conditions, so that’s going to have to change, and I need to re-apply for badge access in my home lab so I can get into the building when I need to go back. I found out the hard way that because the fall term was about to start, my badge stopped working and I had to piss off a few people before I could get into the lab.

Basically, despite already doing one experiment, I still have quite a bit of work to do and not lot of time to do it. That said, it’s back to work I go.