Package mining.extraction

Extract trends and patterns from a set of input streams. More...

Classes
class	KmeansSymbolDistribution
	Create clusters over the distribution of symbols in a set of input streams. More...
class	MaxMiningFunction
	Create a mining function that computes the maximum value of all sequences. More...
class	MaxMiningFunctionProcessor
	Create a mining function from BeepBeep processors. More...

Detailed Description

Extract trends and patterns from a set of input streams.

This is accomplished by an instance of Pat The Miner's SetMiningFunction. This function takes as its input a set of sequences, and returns a "trend" computed on these sequences. A trend can potentially be anything, such an average, a vector of features, a statistical distribution, etc.

Although Pat The Miner provides a few built-in functions (and the possibility for a user to create their own by extending SetMiningFunction), the ProcessorMiningFunction object simplifies the task of creating mining functions through the use of BeepBeep processors. The ProcessorMiningFunction takes as input as set of sequences, and produces as output a "pattern" object, representing a trend computed from the contents of the input sequences.

When created, the ProcessorMiningFunction is parameterized by two BeepBeep Processor objects:

	This processor, called the trend processor, computes a trend from a single input sequence. The trend is taken as the last event output by β when being fed a sequence of events.
	This processor, called the aggregation processor, aggregates the trends computed by β from each input sequence into a single, "aggregated" trend. Its input is an array of values.

In addition, the computation of β's output on each input sequence can be done in parallel in different threads, if one supplies a properly configured ThreadManager to the function.

Depending on how these two parameters are instantiated, the ProcessorMiningFunction processor computes different things. The examples in this section show different ways of using the mining functions, an in particular the ProcessorMiningFunction object.

Author

Sylvain Hallé