6 Replies Latest reply: Jul 23, 2012 11:01 AM by 950757 RSS

    Regression problems

    950757
      Hi,

      i have a problem predicting data in Oracle R Enterprise. I build a model on a part of the dataset and build a prediction on a whole dataset. Something goes wrong - the prediction becomes inadequate. When i build a model on a whole data set everything is fine. I was trying to post the datasets, but they just messed.
      Can anybody help me, please?

      This is the code:

      begin
      sys.rqScriptCreate('electro2',
      'function(dat) {
      mod <- lm(ZHD ~ EL, dat)
      mod}');
      end;
      /
      drop table electro_model;
      create table electro_model as
      select *
      from table(rqTableEval(
      cursor(select ZHD,EL
      from ELECTRISITY where DAT>'01.04.2012' and
      DAT< '01.06.2012'
      ),
      NULL,
      NULL,
      'electro2'));

      -- score data
      begin
      sys.rqScriptCreate('electropred2',
      'function(dat, mod) {
      prd <- predict(mod, newdata=dat)
      prd[as.integer(rownames(prd))] <- prd
      res <- cbind(dat, PRED = prd)
      res}');
      end;
      /
      drop table electro_res
      create table electro_res as
      select *
      from table(rqTableEval(
      cursor(select ZHD,EL from ELECTRISITY),
      cursor(select 1 max1, 1 pos1, 'mod' name1,
      to_number(null) max2, to_number(null) pos2,
      to_char(null) name2, total, chunk, value
      from ELECTRO_MODEL),
      'select ZHD, EL, 1 PRED from ELECTRISITY',
      'electropred2'))
      order by 1, 2, 3;



      Regina Lazareva

      Edited by: 947754 on 20.07.2012 5:47

      Edited by: 947754 on 20.07.2012 5:47

      Edited by: 947754 on 20.07.2012 5:48