1

I need some help with forecasting data in Oracle. I have a set of data in a table called MEASUREMENTS looking like that:

CITY   DATA   VALUE       DATE
C1     T       19.6       2017-10-02 19:01
C1     p       1048.6     2017-10-02 19:01
C1     H       60         2017-10-02 19:01
C2     p       1023.1     2017-10-02 19:01
C2     T       28         2017-10-02 19:01
C2     H       30         2017-10-02 19:01
C1     T       21.6       2017-10-02 20:01
C1     p       1041.6     2017-10-02 20:01
C1     H       50         2017-10-02 20:01
C2     p       1053.1     2017-10-02 20:01
C2     T       28.5       2017-10-02 20:01
C2     H       31         2017-10-02 20:01

Its infinite series of data (at least a lot of data). Created algorithm should forecast VALUE in a given time. To call it: MEASUREMENT(CITY, VALUE, FUTURE_DATE), which should predict the chosen value in the future. I know that Oracle provides forecast method, but I'm stuck. Any one knows or maybe have some example of Oracle forecast using Holt model? Thanks in advance

  • Welcome to Stack Overflow. The Holt-Winters algorithm applies exponential smoothing on the data 3 times, so you'll need to implement exponential smoothing. [See this Wikipedia article](https://en.wikipedia.org/wiki/Exponential_smoothing#Triple_exponential_smoothing), and [this nice write-up by Gregory Trubetskoy](https://grisha.org/blog/2016/01/29/triple-exponential-smoothing-forecasting/). Let us know what problems you encounter implementing this and perhaps someone will be able to help you out. Best of luck. – Bob Jarvis - Слава Україні Nov 22 '18 at 02:27
  • It seems you should use the package [DBMS_DATA_MINING](https://docs.oracle.com/en/database/oracle/oracle-database/18/arpls/DBMS_DATA_MINING.html#GUID-A95A0A38-8A5A-4470-B49F-80D81C588BFC) with `EXSM_MODEL=EXSM_HW` – Marmite Bomber Nov 22 '18 at 06:16

0 Answers0