Hi guys,
I'm trying to run a seasonality test on my time series data using the seasonality function of the PAL library. Unfortunately, I encounter a problem with the definition of the output table called decompose. In the PAL Guide, there are 2 options to definition the decompose table. First one comprises the information of the trend seasonality and the irregular (random) components of a time series. The second only considers the irregular one. As the function indicates to analyze the time series regarding to potential cycles, so I expect to get the cycle component as well. So I tried to perform the test with the first decompose table, but in return I get the following error:
AFL error: Registration of AFLLANG wrapper procedure "TSSEASONAL_PROC" failed with error 'inconsistent parameter description'.
Meaning that the function doesn’t accept the first table…
With the second table, the procedure works fine, but I miss the seasonal component.
Any idea that might solve this problem?
Regards
Chen
Hi Chen,
Could you share your table definition, and I am afraid that there is some problem on it.
Plus, what's your HANA/AFL revision?
Best regards,
Meilin
Hello Meilin,
the type of the decomposition table is definied as
CREATE TYPE PAL_TS_DECOMPOSE_T AS TABLE(
"PERIOD" INTEGER,
"SEASONAL" DOUBLE,
"TREND" DOUBLE,
"RANDOM" DOUBLE
);
I'm not sure about the revision of the AFL Library. I guess it's 121 (under installaed Plug-ins at the administration page in HANA Studio).
Thanks in advance!
Hi Jianlian,
Sorry for the delayed, but such an interface is not available until HANA 2.0. As your HANA version is 1.0 SPS12, you can only use the old interface, i.e., outputs only random component.
Best regards,
Meilin
Hello Meilin,
thank you for the advice. I'm afraid that we don't have the latest AFL revision (121).
The table should be alright, as it's defined as required. Nevertheless here is the table definition
CREATE TYPE SEASONAL_DECOMPOSE_T AS TABLE(
"PERIOD" INTEGER,
"SEASONAL" DOUBLE,
"TREND" DOUBLE,
"RANDOM" DOUBLE
);
Regards
Jian Liang Chen were you able to solve your problem? Kind regards Antoine