1.0 - 8.00 - ARIMA Arguments - 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)
ModelTable
Specify the name of the table where the function outputs the coefficients of the input parameters; that is, the model.
ResidualTable
[Optional] Specify this argument if you intend to input the model to the ARIMAPredict function.

Specify the name of the table where the function outputs the residuals of the input parameters.

TimeColumns
Specify the names of the input_table columns that specify the sequence (time points) of the input parameters. The sequence must have uniform intervals.
TargetColumn
Specify the name of the column that contains the time series data in input_table.
PartitionColumns
[Optional] Specify the columns by which the InputTable is partitioned.
Orders
[Required if you omit OrdersTable, OrderP, OrderD, and OrderQ; disallowed otherwise.] Specify the values of the nonseasonal parameters for the ARIMA model—p, d, and q. p and q must be INTEGER values in the range [0, 10]. d must be the INTEGER 0 or 1.
If you do not specify the values of nonseasonal parameters p, d, and q, the function determines and uses their optimal values.
OrderP
[Required if you omit OrdersTable and Orders; disallowed otherwise.] Specify the value of the nonseasonal parameter p for the ARIMA model, an INTEGER in the range [0, 10].
OrderD
[Required if you omit OrdersTable and Orders; disallowed otherwise.] Specify the value of the nonseasonal parameter d for the ARIMA model, the INTEGER 0 or 1.
OrderQ
[Required if you omit OrdersTable and Orders; disallowed otherwise.] Specify the value of the nonseasonal parameter q for the ARIMA model, an INTEGER in the range [0, 10].
SeasonalOrders
[Optional. Disallowed with SeasonalOrderP, SeasonalOrderD, and SeasonalOrderQ.] Specify the values of the seasonal parameters for the ARIMA model—sp, sd, and sq. sp and sq must be INTEGER values in the range [0, 10]. sd must be an INTEGER in the range [0, 3].
If you specify SeasonalOrders, you must also specify Period.
If you do not specify the values of seasonal parameters sp, sd, and sq, the function does not include seasonal variation in its calculations.
SeasonalOrderP
[Required if you omit OrdersTable and Orders; disallowed otherwise.] Specify the value of the nonseasonal parameter sp for the ARIMA model, an INTEGER in the range [0, 10].
If you specify SeasonalOrderP, you must also specify SeasonalOrderD, SeasonalOrderQ, and Period.
SeasonalOrderD
[Required if you omit OrdersTable and Orders; disallowed otherwise.] Specify the value of the nonseasonal parameter sd for the ARIMA model, an INTEGER in the range [0, 3].
If you specify SeasonalOrderD, you must also specify SeasonalOrderP, SeasonalOrderQ, and Period.
SeasonalOrderQ
[Required if you omit OrdersTable and Orders; disallowed otherwise.] Specify the value of the nonseasonal parameter sq for the ARIMA model, an INTEGER in the range [0, 10].
If you specify SeasonalOrderQ, you must also specify SeasonalOrderP, SeasonalOrderD, and Period.
Period
[Optional] Specify the period of a season (m in the formula). The period must be a positive integer value.
IncludeMean
[Optional. Allowed only with either Orders or all of OrderP, OrderD, and OrderQ.] Specify whether the function adds a mean term to the ARIMA model.
If IncludeMean is 'true', d must be 0 in Orders or OrderD and sd must be 0 in SeasonalOrders or SeasonalOrderD.
Default: 'false'
IncludeDrift
[Optional. Allowed only with either Orders or all of OrderP, OrderD, and OrderQ.] Specify whether the function adds a drift term to the ARIMA model.
If IncludeDrift is 'true', d must be 1 in Orders or OrderD.
Default: 'false'
MaxIterNum
[Optional] Specify the maximum iteration number for estimating the parameters. This value must be a positive integer.
Default: 10000
FitMethod
[Optional] Specify the method for fitting the model parameters:
Option Description
'SSE' (Default) Sum of squared error.
'ML' Maximum likelihood. This method is more accurate, but slower to calculate.