Pass number+Difference between direct score function and hyperbolic (should decrease)+General generalizability+Last highest generalizability (technical bullshit purely for yourself)
Max you are frankly pissing people off, but first of all yourself. Remember in the market the main thing is not to deceive yourself......
Well look Maxim, first of all I had to level classes between 1 and 2, where in the target 0 = 2 cluster.
In the array Mv0[q] you need to feed exactly those inputs, which are indicated by number //10
Attached formatted training file where the first line numbered them in order. Not small you can figure it out. And waiting for a picture in the studio...
Well look Maxim, first of all I had to level classes between 1 and 2, where in the target 0=2 cluster.
In the array Mv0[q] you need to feed exactly those inputs, which indicates the number //10
Attached formatted training file where the first line numbered them in order. Not small you can figure it out. And I'm waiting for a picture in the studio...
Tell me what kind of errors you have
Then I'll send you a more complicated dataset, this one's very simple.
I don't need the code, I have the same thing as Reshetov's, but in python. It gives the mql code right away.
I will have to modify yours for a long time, I'm too lazy.
I'll send you a more complicated dataset later, it's very simple.
I don't need the code, I have the same thing as Reshetov's, but in python. It gives the mql code right away.
I will have to modify yours for a long time, I am too lazy
The errors at the very top are commented out, you pull out the polynomial functions themselves from this file, don't forget the sigmoid and signum functions, signum makes a bundle of two polynomials into a commit, sigmoid is an activation function.
To be honest, when I was running your file I thought I had an error in the calculation of the score, but now unloaded mine and everything fell into place
The second figure should tend to zero, the third to 100 the fourth fixes the third at which the second is the least. That's what looks like a real strategy that works. What you threw Maxim leads to the thought of peeking. It is impossible to get a model of such quality in this area, where unknowable were only 24 vectors and their total number of 550. I have 58 vectors in this picture and I can reach the level of 87% (the third figure) with a pity. The thing is that when you get a sky-high result it gives you pause for thought. So what he is learning it well, Hato the new data ceases to work at all. It is important to work with new data, but not all this.....
Mihail Marchukajtes: The errors at the very top are commented out, you pull out the polynomial functions themselves from this file, don't forget the sigmoid and signum functions, signum makes a bunch of two polynomials in a commit, sigmoid is an activation function.
Well, of course they are not all used. Adding each input doubles the time. My quad-core pulls a maximum of 11 inputs. I optimized your file from top to bottom....
Mihail Marchukajtes: Well, naturally they are not all used. Adding each input doubles the time. My quad-core pulls a maximum of 11 inputs. I optimized your file from top to bottom....
I see, you're an optimizer... I don't understand what you're writing)
Allow the use of cookies to log in to the MQL5.com website.
Please enable the necessary setting in your browser, otherwise you will not be able to log in.
You are signed in. Two-factor authentication has been disabled. Set it up again and scan a new QR code, or continue with two-factor authentication disabled.
Pass number+Difference between direct score function and hyperbolic (should decrease)+General generalizability+Last highest generalizability (technical bullshit purely for yourself)
Max you are frankly pissing people off, but first of all yourself. Remember in the market the main thing is not to deceive yourself......
This is what I do, and that's what I do ...
because this is the data prepared by clustering... everything is already good there
That's exactly what I'm doing.
Well look Maxim, first of all I had to level classes between 1 and 2, where in the target 0 = 2 cluster.
In the array Mv0[q] you need to feed exactly those inputs, which are indicated by number //10
Attached formatted training file where the first line numbered them in order. Not small you can figure it out. And waiting for a picture in the studio...
Well look Maxim, first of all I had to level classes between 1 and 2, where in the target 0=2 cluster.
In the array Mv0[q] you need to feed exactly those inputs, which indicates the number //10
Attached formatted training file where the first line numbered them in order. Not small you can figure it out. And I'm waiting for a picture in the studio...
Tell me what kind of errors you have
Then I'll send you a more complicated dataset, this one's very simple.
I don't need the code, I have the same thing as Reshetov's, but in python. It gives the mql code right away.
I will have to modify yours for a long time, I'm too lazy.
Please tell me how to find an order with the maximum profit (mql4).
Thank you very much.
You tell me what errors you have
I'll send you a more complicated dataset later, it's very simple.
I don't need the code, I have the same thing as Reshetov's, but in python. It gives the mql code right away.
I will have to modify yours for a long time, I am too lazy
To be honest, when I was running your file I thought I had an error in the calculation of the score, but now unloaded mine and everything fell into place
The second figure should tend to zero, the third to 100 the fourth fixes the third at which the second is the least. That's what looks like a real strategy that works. What you threw Maxim leads to the thought of peeking. It is impossible to get a model of such quality in this area, where unknowable were only 24 vectors and their total number of 550. I have 58 vectors in this picture and I can reach the level of 87% (the third figure) with a pity. The thing is that when you get a sky-high result it gives you pause for thought. So what he is learning it well, Hato the new data ceases to work at all. It is important to work with new data, but not all this.....
The errors at the very top are commented out, you pull out the polynomial functions themselves from this file, don't forget the sigmoid and signum functions, signum makes a bunch of two polynomials in a commit, sigmoid is an activation function.
Here's one. There's only 12 fiches, not 24.
I have errors
>>> print(train_score, " ", tst_score)
1.0 0.5454545454545454
Here's one. There are only 12 fiches, not 24
I have errors
>>> print(train_score, " ", tst_score)
1.0 0.5454545454545454
Well, naturally they are not all used. Adding each input doubles the time. My quad-core pulls a maximum of 11 inputs. I optimized your file from top to bottom....
I see, you're an optimizer... I don't understand what you're writing)