SVMSparse Arguments - Teradata Vantage

Machine Learning Engine Analytic Function Reference

Product

Teradata Vantage

Release Number

8.00

1.0

Published

May 2019

Language

English (United States)

Last Update

2019-11-22

dita:mapPath

blj1506016597986.ditamap

dita:ditavalPath

blj1506016597986.ditaval

dita:id

B700-4003

lifecycle

Product Category

Teradata Vantage™

ModelTable

Specify the name of the model table (which must not exist).

IDColumn

Specify the name of the input_table column that contains the identifiers of the training samples.

AttributeNameColumn

Specify the name of the input_table column that contains the attributes of the samples.

AttributeValueColumn

[Optional] Specify the name of the input_table column that contains the attribute values.

Default behavior: Each attribute has the value 1.

ResponseColumn

Specify the name of the input_table column that contains the classes of the samples.

Cost

[Optional] Specify the regularization parameter λ in the SVM soft-margin loss function:

The cost must be greater than 0.0.

Default: 1.0

Bias

[Optional] Specify whether to add another dimension containing the bias value b. The bias must be nonnegative. If bias is greater than 0, the function converts each sample

in the training set to (

, b). Use this argument when not all samples center at 0.

Default: 0.0

HashProjection

[Optional] Specify whether to use hash projection on attributes. Hash projection can accelerate processing speed but can slightly decrease accuracy.

You must use hash projection if the data set has more features than fit into memory.

Default: 'false'

HashBuckets

[Optional] [Valid only with HashProjection ('true').] Specify the number of buckets for hash projection. Usually, the function can determine the appropriate number of buckets from the scale of the input data set. However, if the data set has many features, you might have to specify buckets_number to accelerate the function.

Default behavior: The function determines the appropriate number of buckets from the scale of the input data set.

ClassWeights

[Optional] Specify the weights for different classes. If you specify a weight for a class, the cost parameter for that class is weight * cost. A weight larger than 1 often increases the accuracy of class; however, it may decrease global accuracy.

Default behavior: The function assigns weight 1.0 to any class not assigned a weight in this argument.

MaxStep

[Optional] Specify the maximum number of steps of the training process. One step means that the trainer sees each sample once. The max_step must be in the range (0, 10000].

Default: 100

Epsilon

[Optional] Specify the termination criterion: When the difference between the values of the loss function in two sequential iterations is less than this epsilon, the function stops. The epsilon must be greater than 0.0.

Default: 0.01

Seed

[Optional] Specify the random seed the algorithm uses for repeatable results (for more information, see Nondeterministic Results). The algorithm uses the seed to order the training set randomly and consistently. The seed must be a nonnegative LONG value.

Default: 0