From: dataminer101@yahoo.com   
      
   Hi Ted and thanks for the reply.   
      
   As I told another person who kindly replied to my post, Since the   
   produced model will be used as part of an alarm/flagging system, I will   
   have to produce a curve of each of the parameters of interest using 4   
   values/day=once/6h, and do this for the 44 days, this is to flag and   
   correct any abnormal behaviour ASAP.   So, the whole curve would have   
   4*44=176 values. E.g. for the water consumption curve: day1:   
   12AM=65Gal, 6AM=150, 12PM ... DAY44=6PM=1500Gal. I would have to come   
   up with similar curves for each of the parameters of interest   
   (inputs/outputs).  Now as far as ANNs are concerned, do I have to   
   produce 176 of these ANNs, one for each predicted value?  ANN1: input1   
   (temperature-value Day1@12AM) input2 (humidity-value Day1@12AM)...   
   output1 (feed consumption-value Day1@12AM), output2   
   (heater_runtime-values Day1@12AM)... and train the ANN with the 50-60   
   samples (Day1@12AM) from previous productions. This would produce an   
   ANN for predicting the value of each parameter for Day1@12AM for future   
   productions, etc.... This would quite intensive computationally, so I   
   am wondering if there is a better way to maybe feed-in all the 176   
   values time series in one shot to have something like   
   input1(temperature-values 1-176), input2(humidity-values 1-176)...   
   output1(feed consumption-values 1-176), output2 (heater runtime-values   
   1-175)...  and this will produce only one ANN which will predict the   
   176 values for all parameters of future productions?   
   I would really appreciate your help as I am really stuck at this.   
      
   [ comp.ai is moderated.  To submit, just post and be patient, or if ]   
   [ that fails mail your article to , and ]   
   [ ask your news administrator to fix the problems with your system. ]   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)