TS_LACK OF FIT function [Time Series]

Performs the lack-of-fit test for a univariate time series or transfer function, given the appropriate correlation function.

TS_LACK OF FIT (timeseries_expression, p_value, q_value, lagmax, [tolerance])

OVER (window-spec)

This function returns a double-precision floating-point value containing the lack-of-fit statistic (q) for the time series. TS_LACK_OF_FIT calls the function imsls_d_lack_of_fit in the IMSL libraries.

The arguments of TS_LACK_OF_FIT map to the IMSL library function imsls_d_lack_of_fit as follows:

params = imsls_d_arma(n_objs, z[], p, q,	IMSLS_LEAST_SQUARES,	IMSLS_CONVERGENCE_TOLERANCE, tolerance,	IMSL_RESIDUAL, &residual, 0);correlations = imsls_d_autocorrelation(n_objs-p+lagmax, residuals, lagmax, 0);result = imsls_d_lack_of_fit(n_objs, correlations, lagmax, npfree, 0);

For detailed information on how the IMSL function imsls_d_lack_of_fit performs time series calculations, see IMSL Numerical Library User’s Guide: Volume 2 of 2 C Stat Library.

This example shows an input data table, a SQL statement containing the TS_LACK_OF_FIT function, and the data values returned by the function. This example uses the following table (called DATASET) as its input data. The DATASET table contains 50 rows of time series data:

The following SQL statement returns the lack of fit statistic on data from the data column:

select ts_lack_of_fit(data,1,1,5,0.225) over (order by rownum rows between unbounded preceding and unbounded following) as res from DATASET

Sybase IQ returns 50 rows, each containing the same value: