classifying accelerometer data over time in python

Question

I have a set of accelerometer data recordings after 4 different kinds of events (and many repetitions of each). each point (recording) is defined by 3 vectors (X,Y,Z) with 150,000 samples each (5 seconds with a sample rate of 30KHz). so all in all i have alot of 3x150,000 matrices.

I want to use a classifier or clustering for these points using python in order to see if there is a difference between the behavior of the accelerometer after the different types of events.

1. what is the best classifier to use?
2. how should i structure my data?
3. how do i plot this in order to visualize my data?

thanks a lot to any helpers.

Answer 1

It is quite difficult to give you exact pointers as to what you should be doing in terms of data storage. I suppose you have figured this out already? If not it all depends on various parameters, such as the frequency of read/write access to it, the size of the data and other things, such as your DAQ system. For that matter, I assume this is already resolved and the data is already recorded and stored somehow. This can be in a csv file, a database or a custom file format. As long as you can read it, it doesn't really matter.

Use pandas to structure your data. It is a wrapper around numpy and is described as:

pandas is an open source, BSD-licensed library providing high-performance, easy-to-use data structures and data analysis tools for the Python programming language.

For plotting, since you seem to be doing statistical analysis, either use matplotlib for your basic plotting, or seaborn for your clustering and statistical analysis etc. From the Seaborn website:

Seaborn is a Python visualization library based on matplotlib. It provides a high-level interface for drawing attractive statistical graphics.

Have a look at the Seaborn gallery , for possible plots. It gives a good overview on the diagrams that are possible with the library as well as the source code that is required for that plot.