Time Series Analysis Package
UNBIASEDNESS Method
rc = TSA.UNBIASEDNESS (y, predict, <siglevel>, intercept, scale, fvalue, pvalue);
The UNBIASEDNESS function tests whether a univariate time series is unbiased.
Required Arguments
You must specify the following arguments, separated by a comma:
- y
specifies the input time series array.
- predict
specifies an input array of predicted time series.
Optional Arguments
You can also specify the following argument. If you want to use a default value for this argument, enter a space for it.
- siglevel
specifies the significance level.
Returned Values
The UNBIASEDNESS function returns the following values:
- rc
-
returns one of the following scalar return codes:
rc Termination Reason 1 Biased predictions 0 Unbiased predictions –1 Degree of freedom error –2 Singular system –3 Extreme value - intercept
returns the constant parameter.
- scale
returns the scale parameter.
- fvalue
returns the test statistic for the F test.
- pvalue
returns the p-value for the F test.
Example
This example uses the TSMODEL procedure to test whether the series Actual is unbiased:
proc hpf data=sashelp.air out=_null_ outfor=outfor;
id date interval=month;
forecast air;
run;
proc reg data=outfor;
model actual=predict;
test intercept=0, predict=1;
run;
quit;
data mylib.outfor;
set outfor;
run;
proc tsmodel data=mylib.outfor outscalar=mylib.bias_scalar outarray=mylib.bias_array;
id date interval=month;
var ACTUAL PREDICT;
outscalars intercept scale fvalue pvalue;
require tsa;
submit;
declare object TSA(tsa);
rc=TSA.UNBIASEDNESS(ACTUAL, PREDICT, 0.05, intercept, scale, fvalue, pvalue);
endsubmit;
run;