1.0 - 8.00 - FellegiSunter Example 2: Supervised Learning - Teradata Vantage

Teradata® Vantage Machine Learning Engine Analytic Function Reference

Product
Teradata Vantage
Release Number
1.0
8.00
Release Date
May 2019
Content Type
Programming Reference
Publication ID
B700-4003-098K
Language
English (United States)

The unsupervised model is created like the supervised model, except the argument TagColumn ('match_tag') specifies the data for the on which to train the model. Supervised learning does not use initialization parameters.

Input

SQL Call

CREATE MULTISET TABLE "fg_supervised_model" AS (
  SELECT * FROM FellegiSunter (
    ON fstrainer_input AS InputTable
    USING
    ComparisonFields ('jaro1_sim: 0.8', 'ld1_sim:0.8', 'ngram1_sim:0.5',
                      'jw1_sim:0.8')
    TagColumn ('match_tag')
  ) AS dt
) WITH DATA;

Output

This query returns the following table:

SELECT * FROM fg_supervised_model ORDER BY 1;
fg_supervised_model
_key _value
comparisoncnt 4
comparisonname_0 jaro1_sim
comparisonname_1 ld1_sim
comparisonname_2 ngram1_sim
comparisonname_3 jw1_sim
comparisonthreshold_0 0.8
comparisonthreshold_1 0.8
comparisonthreshold_2 0.5
comparisonthreshold_3 0.8
is_supervised true
lambda 0.9
lower_bound -0.415037499278844
m_0 0.9999999
m_1 0.166666666666667
m_2 0.5
m_3 0.9999999
mu 0.9
time_used 35.413000 seconds
u_0 0.666666666666667
u_1 1.0E-7
u_2 1.0E-7
u_3 0.833333333333333
upper_bound -0.415037499278844