avoid the need to predetermine vector size by simply picking a reasonable size and shoehorning the training data into vectors of that size
Used in SGD

\n

\n

shoehorning

\n

picking one or more locations by using

\n

> a hash of the name of the variable for continuous variables

\n

> a hash of the variable name and the category name or word for categorical, text-like, or word-like data

\n

\n

\n

ENCODING CONTINUOUS FEATURES:

\n

ContinuousValueEncoder : updates one location in the vector

\n

setProbes() - how many locations are to be updated

\n

Vector v = new RandomAccessSparseVector(NUMBER_OF_FEATURES).

\n

If value is String, encoder.addToVector(value, vector)

\n

If value is Double, encoder.addToVector(null, value, vector)

\n

\n

ENCODING CATEGORICAL AND WORD-LIKE FEATURES:

\n

WordValueEncoder: the default number of probes is two and the location being updated varies according to the value specified as well as the name of the variable

\n

1. AdaptiveWordValueEncoder:

\n

2. StaticWordValueEncoder

\n

\n

ENCODING TEXT-LIKE FEATURES:

\n

Encoding a text-like feature is similar to encoding a word-like one except that text has many words.

\n

Mahout Classification