Stack Exchange
log in

pat

 Discussion between Motig and pat

Imported from a comment discussion on stats.stackexchange.com/...
pat
Jan 25, 2013 23:36
You could also just force the output to +/-1 levels by using an ifelse threshold.
pat
Jan 25, 2013 23:35
Good Luck
pat
Jan 25, 2013 23:35
see if that dynamic range increase helps any
pat
Jan 25, 2013 23:34
I'd consider to expand it by scaling to +/-1
pat
Jan 25, 2013 23:34
what I'm seeing is the output data is strongly bunched up betwee +/-.03
pat
Jan 25, 2013 23:34
so hope that adds to your confidence-)
pat
Jan 25, 2013 23:34
Last thing I'll pint out is that I trained a nnet in R and also got zero
pat
Jan 25, 2013 23:33
R is a statistical programming language.
pat
Jan 25, 2013 23:27
I'm trying to run a quick nn model here in R. If you want to wait.. it's ok, otherwise I'd appreciate a thread vote if you found anything useful.
pat
Jan 25, 2013 23:23
ok input look ok output is a llittle skewed (min,max = -.49 to 1)
pat
Jan 25, 2013 23:23
nput18 Input19 Output0
Min. :-1.0000 Min. :-1.0000 Min. :-0.4943881
1st Qu.:-0.3515 1st Qu.:-0.3515 1st Qu.:-0.0344828
Median :-0.2361 Median :-0.2361 Median : 0.0009466
Mean :-0.2410 Mean :-0.2410 Mean : 0.0007731
3rd Qu.:-0.1319 3rd Qu.:-0.1319 3rd Qu.: 0.0355646
Max. : 1.0000 Max. : 1.0000 Max. : 1.0000000
pat
Jan 25, 2013 23:18
ah.. i just realized why i'm having so many problems. The data has commas instead of decimals! I need to preprocess.
pat
Jan 25, 2013 23:07
still working on processing data, i assume target is col 21. takes a while in R.
pat
Jan 25, 2013 22:58
that is better. hang on.
pat
Jan 25, 2013 22:57
And maybe you could limit it to just 500 observations. It's taking a while to download.
pat
Jan 25, 2013 22:56
Also, your .dat output has data on seperate columns making it hard to process. It would be nice to have each row correpsond to each input/output observation.
pat
Jan 25, 2013 22:53
I'm confused. The table format should be input D={x1,x2,...xj} where x is your input variable columns and Y=Yt the target variable. Forget slope, just use training data that makes sense. Can you get it in this format?
pat
Jan 25, 2013 22:49
That is odd. I would start by making sure that there is ONLY ONE target variable. Get rid of unnecessary ones.
pat
Jan 25, 2013 22:48
?
pat
Jan 25, 2013 22:46
which column is the target variable? The last column?
pat
Jan 25, 2013 22:43
Your data has no labels for input and output.
pat
Jan 25, 2013 22:43
Hi, I've never used chat... so sorry for the delay, but this is a better idea.
pat
Jan 25, 2013 22:39
Not price, but return as i mentioned. Price is explosive and not a good variable to use for NNs. You want to always train any type of learner with stationary data where possible. You can always transform back to price after. I have a feeling we are cluttering up the thread with conversation; any suggestions on how to limit that would be appreciated.
pat
Jan 25, 2013 22:39
Unfortunately, It doesn't help the troubleshooting much with only raw data. It would be more useful to post the transformed training data (with input and target variables specified).
pat
Jan 25, 2013 22:39
I think you mean the tanh function. Could you show a table of some of the training data? Also, what happens if you use the default output (not slope), is the data non-zero? I'm not familiar with the Sierra Charts tool, so there may be something tool specific I might miss.