Day 29, July 1 – Vidushi

Contrary to the frustration of yesterday, I feel like I finally got the reap the benefits of rewriting code today! I was beginning to wonder if it was a pointless task, but I think I understand how the code works much, much better than I did, and I had a couple of important realizations that help solve my problems from a few days ago. Specifically, these were as follows:

  • One of the issues I was having was that when comparing one set of accelerations to another, I was finding that one set was smaller than the other; a big proportion of the accelerations in one of the sets just didn’t exist in the other. Today, I *think* this is because one set was a set of average accelerations per word, and the other is just all the acceleration values recorded. To fix this, I got rid of the averages altogether; I didn’t end up using that information anyway.
  • A second thing is that in my mass dictionary that mapped each word to accelerations recorded for it, I didn’t sort this by the text. For example, if the word “the” appears in Text 1 with an acceleration of x, and in Text 4 with acceleration of y, then my data structure did not distinguish between the two; it just stored it as as { “the”: [x, y] }. Obviously, this makes no sense, so I changed my data structure entirely, and now it knows which word goes with which text.

In this new data structure I mentioned in the second point, the fundamental way of parsing is so different that the work I had done to label points is slightly obsolete now. The rest of the program has been successfully replicated. This means that tomorrow I need to fix the labelling of the points, then restart the classification of words based on what change in acceleration they cause. Basically, I am back at the point where I started last week, but with a much better understanding of what the program is doing, and a much higher confidence level that I can actually make the changes I want to make. Hopefully, then, tomorrow I can implement the labels properly, assign the word classification, and get started on all the other data analysis results we want to see.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s