Exercises

Try to work through the following exercises:

• Use accessSplit.py to learn how the sliding window size affects the construction speed of the 2M.gz time series from Chapter 4. Perform your experiments for the following sliding window sizes: 16, 256, 1024, 4096, 16384, and 32786.
• Can you resolve the overflow situations with accessSplit.py and the 500.gz time series we came across at the beginning of the chapter?
• Try reducing the threshold values in the speed.py examples presented in the Checking the search speed of iSAX indexes section and see what happens.
• Create two time series with 250,000 elements each and use speed.py to understand their behavior when the number of segments is in the 20 to 40 range. Do not forget to use an appropriate sliding window size.
• Experiment with speed.py but this time, change the threshold value instead of the number of segments. Is the threshold value more important than the number of segments in the search speed of an iSAX...