Community Profile

Mehmet Demirel


University of Twente

13 total contributions since 2009

Contact

Contributions in
View by

Posted


setx and newff.. two networks, the same setx parameters but different sim results..
Hello, I test two cases, 1) create a network using nntool and train it and simulate for a test data Result1 = sim(network1,IN)...

4 years ago | 1316 views

Responded


Re: area and jbfill codes
Yes this is exactly what i need!! Thanks Sadik!! Regards, Mehmet

4 years ago | 2716 views

Posted


area and jbfill codes
Hello, I was using area function to draw grey shaded confidence interval (5%-95%). X=[Q95 Q5]; area(X) where Q95 and Q5 are ...

4 years ago | 2716 views

Posted


trainbr and trainlm: why different performance figures?
Hello, I am testing two training functions using nntool and two versions of Matlab: R2008b and R2010b. I find it very strange w...

5 years ago | 2503 views

Responded


Re: Objective function for newff model with trainlm
This is an interesting approach if i can figure out more. I have several unclear parts as listed below, //what do you mean by ...

6 years ago | 5379 views

Responded


Re: Objective function for newff model with trainlm
I will try to explain better; we intend to forecast low flows (water discharge below 1100 m3s-1) in the Rhine River, using relev...

6 years ago | 5379 views

Responded


Re: Objective function for newff model with trainlm
Thanks for the quick reaction, I have red this and already got into my code MSEgoal = 0.01*mean(var(t')) I looked now at "heath ...

6 years ago | 5379 views

Posted


Objective function for newff model with trainlm
Hello, I am confused with introducing a new objective function to newff network. I use trainlm supervised training method Afte...

6 years ago | 5379 views

Responded


Re: Xtick and Xticklabel handeling.
Hello Serval, I tried to use your code for an other y data. However the Y axis has always 0:0.1:1 values overwritten on my real...

6 years ago | 16529 views

Responded


Re: ARMAX Parameters (using without Matlab)
q is a shift operator, see below for a simple example, Armax Model: y(t) = B(q)u(t) + C(q)e(t) B(q) = 0.7354 (+-0.01128) q^-7 ...

8 years ago | 2528 views

Posted


Prediction with ARMAX
Question is about data introduction to armax: M is a two column matrix containing output (y) and input (u) time series. Each row...

8 years ago | 7071 views

Posted


Levenberg-Marquardt (trainlm) and Nash-Sutcliffe efficiency coefficient
After training a feedforward network with trainlm a fast convergence is reached with my data. c=Getx(net) will get all network w...

8 years ago | 4864 views

Responded


Re: Trainlm error definition
Dear Pavan And Greg, I have a similar but simplier problem in performance function: I would like to edit mse.m because it calc...

8 years ago | 3318 views