Time Series Indexing: Implement iSAX in Python to index time series with confidence

Cover
Title Page
Copyright
Dedication
Contributors
Table of Contents
Preface
Chapter 1: An Introduction to Time Series and the Required Python Knowledge
	Technical requirements
	Understanding time series
		Time series are everywhere
		Essential definitions
		Time series data mining
		Comparing time series
		The Euclidean distance
		The Chebyshev distance
	What is an index and why do we need indexing?
	The Python knowledge that we are going to need
		Timing Python code
		An introduction to Anaconda
		The required Python packages
		Setting up our environment
		Printing package versions
		Creating sample data
		Publicly available time series data
		How time series are processed
	Reading time series from disk
		Is all data numeric?
		Do all lines have the same amount of data?
		Creating subsequences
	Visualizing time series
	Working with the Matrix Profile
	Exploring the MPdist distance
	Summary
	Resources and useful links
	Exercises
Chapter 2: Implementing SAX
	Technical requirements
	The required theory
		Why do we need SAX?
		Normalization
		Visualizing normalized time series
	An introduction to SAX
		The cardinality parameter
		The segments parameter
		How to manually find the SAX representation of a subsequence
		Ηow can we divide 10 data points into 3 segments?
		Reducing the cardinality of a SAX representation
	Developing a Python package
		The basics of Python packages
		The SAX Python package
	Working with the SAX package
		Computing the SAX representations of the subsequences of a time series
	Counting the SAX representations of a time series
	The tsfresh Python package
	Creating a histogram of a time series
	Calculating the percentiles of a time series
	Summary
	Useful links
	Exercises
Chapter 3: iSAX – The Required Theory
	Technical requirements
	Background information
		Trees and binary trees
	Understanding how iSAX works
		The cardinality parameter
		The segments parameter
		The threshold parameter
		Computing the normalized mean values
		How big can an iSAX index get?
		What happens when there is no space left for adding more subsequences to an iSAX index?
	How iSAX is constructed
		How iSAX is searched
		Promotion strategy
		Splitting nodes
	Manually constructing an iSAX index
	Updating the counting.py utility
	Summary
	Useful links
	Exercises
Chapter 4: iSAX – The Implementation
	Technical requirements
	A quick look at the iSAX Python package
		The class for storing subsequences
		The class for iSAX nodes
		The class for entire iSAX indexes
	Explaining the missing parts
	Exploring the remaining files
		The tools.py file
		The variables.py file
		The sax.py file
	Using the iSAX Python package
		Reading the iSAX parameters
		How to process subsequences to create an iSAX index
		Creating our first iSAX index
		Counting the subsequences of an iSAX index
		How long does it take to create an iSAX index?
		Dealing with iSAX overflows
	Summary
	Useful links
	Exercises
Chapter 5: Joining and Comparing iSAX Indexes
	Technical requirements
	How the sliding window size affects the iSAX construction speed
	Checking the search speed of iSAX indexes
	Joining iSAX indexes
	Implementing the joining of iSAX indexes
	Explaining the Python code
	Using the Python code
		We have a long list of Euclidean distances, so what?
		Saving the output
		Finding iSAX nodes without a match
	Writing Python tests
		What are we going to test?
		Comparing the number of subsequences
		Checking the number of node splits
		All Euclidean distances are 0
		Running the tests
	Summary
	Useful links
	Exercises
Chapter 6: Visualizing iSAX Indexes
	Technical requirements
	Storing an iSAX index in JSON format
		Downloading the JavaScript code locally
		Running the code locally
	Visualizing an iSAX index
		Visualizing iSAX as a tree
	Trying something radical
	More iSAX index visualizations
	Using icicle plots
	Visualizing iSAX as a Collapsible Tree
	Summary
	Useful links
	Exercises
Chapter 7: Using iSAX to Approximate MPdist
	Technical requirements
	Understanding the Matrix Profile
		What does the Matrix Profile compute?
		Manually computing the exact Matrix Profile
	Computing the Matrix Profile using iSAX
		What happens if there is not a valid match?
		Calculating the error
		Approximate Matrix Profile implementation
		Comparing the accuracy of two different parameter sets
	Understanding MPdist
		How to compute MPdist
		Manually computing MPdist
	Calculating MPdist using iSAX
	Implementing the MPdist calculation in Python
		Using the approximate Matrix Profile way
		Using the join of two iSAX indexes
	Using the Python code
		Comparing the accuracy and the speed of the methods
	Summary
	Useful links
	Exercises
Chapter 8: Conclusions and Next Steps
	Concluding all that we have learned so far
	Other variations of iSAX
	Interesting research papers on time series
	Interesting research papers on databases
	Useful books
		Useful books on databases
		Building a strong computer science background
		Books on UNIX and Linux
		Books on the Python programming language
	Summary
	Useful links
	Exercises
Index
About Packt
Other Books You May Enjoy