Discussion of article "Third Generation Neural Networks: Deep Networks" - page 13

New comment
 
jake89:

Hi

Also , following code :  

I am not sure I understand it . What happens if I have a new X vector and I want to preprocess it  and run pr.sae<-nn.predict(SAE, X); 

How do I do it ? Thank you .

newX <- predict(spSign, X)
pr.sae <- nn.predict(SAE, newXX)
# Calculate parameters preprocessing
spSign <- preProcess(x[t$tr, ], method = "spatialSign")
# Using these parameters (spSign) carry out the actual preprocessing 
x.tr<-predict(spSign, x[t$tr, ])
# Using these parameters (spSign) carry out the actual preprocessing  
x.ts<-predict(spSign, x[t$ts, ]

Description of the function preProcess(), refer to the package "caret".

Best regards


 
Vladimir Perervenko:

Description of the function preProcess(), refer to the package "caret".

Best regards


I decided to just use your code ... But I am stuck on the "No calculation results! Symbol" error .

I see in the code , a server with port is referenced . What server is this referring to ?  

 
jake89:

I decided to just use your code ... But I am stuck on the "No calculation results! Symbol" error .

I see in the code , a server with port is referenced . What server is this referring to ?  

Hi,

What do you run that carried out?

I can not read minds at a distance.

Please describe your problem more detail.

Best regards

Vlad

 
Vladimir Perervenko:

Hi,

What do you run that carried out?

I can not read minds at a distance.

Please describe your problem more detail.

Best regards

Vlad

ok sorry . I will see what else I can figure out . I get the "No calculation results! Symbol" and I install the indicator and still get the error .

I made some changes but markets is closed right now . Will let you know next week . 

 
jake89:

ok sorry . I will see what else I can figure out . I get the "No calculation results! Symbol" and I install the indicator and still get the error .

I made some changes but markets is closed right now . Will let you know next week . 

Hi,

The problem appeared after the release of a new version of the svSocket () package.

I have not found the cause of the data block between the client and the server.

I rewrote the expert, and attached it to a new article which is to be released a few days ago (today at checkout).

Best regards

Vladimir

 

Rterm crashed!

Rterm crashed!

Rterm crashed!

Rterm crashed!

 
The most efficient way to manage this is through the Windows Task Manager. When the EA or indicator is loaded, if the Rterm does not appear in the task list, then the R processor has crashed. The main cause of this problem is caused by a syntax error in the script, where the length of the received MQL vectors does not match the length of the vectors parsed from the Rterm.

This problem can be fixed by debugging the script line by line from beginning to end in Rstudio
 

So, after a long debugging and monitoring of the work turned out to be this.

I refined the script to work in strategy testing (it takes a lot of time for testing!).

Moved everything from the OnTimer() function to action(), added the OnTick() function. I added the timer_enable = true/false option and the switch_count_ticks variable. The result is approximately as follows:

 void OnTimer()
{
   if(timer_enable)
    {
      action();
    }
}
void OnTick()
{
   count_ticks++;
   if(sig == 0  || op == "WAIT")
   {
      CheckForClose(op, magic, sig);
   }

   if(timer_enable) return;
   if(count_ticks >= switch_count_ticks)
   {
      count_ticks=0;
      if(!timer_enable)
      {
         action();
      }
   }
   //action();
}

In the tester we select timer_enable = false and set switch_count_ticks = 200. The value turned out to be optimal for me to test at least a week in a reasonable time. We leave the tester speed as default.

The best results were recorded before the opening of sessions and a short time after. Night time was switched off.

 
Insert the code correctly, please. I've corrected it
 
kimkarus:

So, after a long debugging and monitoring of the work turned out to be this.

I refined the script to work in strategy testing (it takes a lot of time for testing!).

Moved everything from the OnTimer() function to action(), added the OnTick() function. I added the timer_enable = true/false option and the switch_count_ticks variable. The result is approximately the following:


In the tester we select timer_enable = false and set switch_count_ticks = 200. The value turned out to be optimal for me to test at least a week in a reasonable time. We leave the tester speed as default.

The best results were recorded before the opening of sessions and a short time after. Night time was switched off.

Good afternoon.

What script are we talking about?

Could you describe in more detail what is in the script?

I understand you managed to run the script with the R process in the tester?

If so, that's interesting.

Please take your time and describe it in as much detail as possible. Is the R process executed in a client-server bundle or in a single Rterm?

1...67891011121314151617
New comment