VARMAX Example 2: Exogenous Model - 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
previous
Product Category
Teradata Vantageā„¢

This example models the expenditure time series as a function of exogenous variables (income and investment) and the AR and MA parameters. The function chooses the order and lag of the variables according to the degree of complexity, typically a trade-off between accuracy and over-fitting.

Input

SQL Call

SELECT * FROM VARMAX (
  ON finance_data3 AS "data" PARTITION BY id ORDER BY period
  USING
  ResponseColumns ('expenditure')
  ExogenousColumns ('income', 'investment')
  PartitionColumns ('id')
  PDQ ('1, 1, 1')
  Lag (3)
  ExogenousOrder (3)
  IncludeMean ('false')
  StepAhead (3)
) AS dt ORDER BY id;

Output

The model converges for all three decades.

id coef coef_value stepahead predict_expenditure
1 coef 1, 1, 1, 0, 0, 0, 3    
1 ar_params [[-0.11048447321369022]]    
1 ma_params [[-0.1925561123625604]]    
1 exogenous_params [[0.15184265006071107, -0.6548579365873168]], [[3.4356161960447644, -3.8815455917944566]], [[1.352629223976695, 1.6692441091760393]]    
1 seasonal_ar_params      
1 seasonal_ma_params      
1 period 0    
1 lag 3    
1 sigma [[21815.169519750183]]    
1 aic 10.400617266572384    
1 bic 10.74186068116308    
1 iterations 18    
1 converged true    
1     1 4601.67249271766
1     2 6375.70446779608
1     3 6353.20194246941
2 coef 1, 1, 1, 0, 0, 0, 3    
2 ar_params [[0.7135109382825823]]    
2 ma_params [[0.8116168306474483]]    
2 exogenous_params [[-1.3199094913183353, 3.7663680262105013]], [[0.8573542123886598, -2.361564228588058]], [[0.7742458737629424, -2.39596174873262]]    
2 seasonal_ar_params      
2 seasonal_ma_params      
2 period 0    
2 lag 3    
2 sigma [[41347.85268103083]]    
2 aic 11.040032179034085    
2 bic 11.381275593624782    
2 iterations 18    
2 converged true    
2     1 1738.29815872784
2     2 1683.52303336657
2     3 1737.17454282084
3 coef 1, 1, 1, 0, 0, 0, 3    
3 ar_params [[0.7034293980424465]]    
3 ma_params [[-0.5320720491694714]]    
3 exogenous_params [[0.21057556006931047, -0.5831916151232963]], [[-0.09419582311809366, 0.2501649697569839]], [[0.4552695851877213, -1.121807842101149]]    
3 seasonal_ar_params      
3 seasonal_ma_params      
3 period 0    
3 lag 3    
3 sigma [[1.2480831335136702E-25]]    
3 aic -55.88847298918466    
3 bic -55.59909460896767    
3 iterations 25    
3 converged true    
3     1 2321.97511577896
3     2 2404.77038121072
3     3 2433.88438933071