A histogram of (value, frequency) elements, assumed to be sorted in descending order of frequency.
Generate a histogram extractor that maps a sequence of type V to a vector where the position corresponding to each value present in the sequence is the frequency of that value in the sequence.
Generate a histogram extractor that maps a sequence of type V to a vector where the position corresponding to each value present in the sequence is the frequency of that value in the sequence.
a string prepended to feature names. The name of feature corresponding to value v is namePrefix + v.toString.
a "virtual" value corresponding to any value that is not in the histogram. If not set, then undefined values result in an all-zero output vector.
If set, then histogram values with frequency < minFreq will not be defined in the extractor mapping.
If set, then histogram values with frequency > maxFreq will not be defined in the extractor mapping.
If set, then histogram values with probability < minProb will not be defined in the extractor mapping.
If set, then histogram values with probability > maxProb will not be defined in the extractor mapping.
If set, then the subsequence of histogram values that pass any of the above filters is clipped to maxSize elements, and the remaining elements are the ones defined in the extractor mapping.
An extractor that implements a histogram of the values in an input sequence.
the histogram filters above are applied in the following order: minFreq, maxFreq, minProb, maxProb, maxSize.
A histogram of (value, frequency) elements, assumed to be sorted in descending order of frequency.
Generate a "multi-hot" extractor that maps a sequence of type V to a vector where the position corresponding to each value present in the sequence is 1, and all others elements are 0.
Generate a "multi-hot" extractor that maps a sequence of type V to a vector where the position corresponding to each value present in the sequence is 1, and all others elements are 0.
a string prepended to feature names. The name of feature corresponding to value v is namePrefix + v.toString.
a "virtual" value corresponding to any value that is not in the histogram. If not set, then undefined values result in an all-zero output vector.
If set, then histogram values with frequency < minFreq will not be defined in the extractor mapping.
If set, then histogram values with frequency > maxFreq will not be defined in the extractor mapping.
If set, then histogram values with probability < minProb will not be defined in the extractor mapping.
If set, then histogram values with probability > maxProb will not be defined in the extractor mapping.
If set, then the subsequence of histogram values that pass any of the above filters is clipped to maxSize elements, and the remaining elements are the ones defined in the extractor mapping.
An extractor that implements a multi-hot encoding of values in an input sequence.
the histogram filters above are applied in the following order: minFreq, maxFreq, minProb, maxProb, maxSize.
Generate a "one-hot" extractor that maps a value of type V to a vector where the position corresponding to that value is 1, and all others elements are 0.
Generate a "one-hot" extractor that maps a value of type V to a vector where the position corresponding to that value is 1, and all others elements are 0.
a string prepended to feature names. The name of feature corresponding to value v is namePrefix + v.toString.
a "virtual" value corresponding to any value that is not in the histogram. If not set, then undefined values result in an all-zero output vector.
If set, then histogram values with frequency < minFreq will not be defined in the extractor mapping.
If set, then histogram values with frequency > maxFreq will not be defined in the extractor mapping.
If set, then histogram values with probability < minProb will not be defined in the extractor mapping.
If set, then histogram values with probability > maxProb will not be defined in the extractor mapping.
If set, then the subsequence of histogram values that pass any of the above filters is clipped to maxSize elements, and the remaining elements are the ones defined in the extractor mapping.
An extractor that implements a one-hot encoding of input values.
the histogram filters above are applied in the following order: minFreq, maxFreq, minProb, maxProb, maxSize.
A model for generating Extractor objects from a histogram of values
The value type of elements in histogram
A histogram of (value, frequency) elements, assumed to be sorted in descending order of frequency.