Machine learning in trading: theory, models, practice and algo-trading - page 1992

[Maxim Dmitrievsky]

mytarmailS:

I'm talking about python, I can't run your own code in this miracle language for clean guys.)

I'll write in python, but the code must first run))

Try running it through the debugger first.

It'll create a config file for you

Then try just running

[Evgeniy Chumakov]

Hello all!

If anyone has free time, please check the time series for predictability by neural networks.

As I do not understand anything in these ns, and there is no one else to ask... Thank you!

This is what the series looks like (attached)

The last time value is at the beginning of the file, the old one is at the end.

[Farkhat Guzairov]

Evgeniy Chumakov:

Hello all!

If anyone has free time, please check the time series for predictability by neural networks.

As I do not understand anything in these ns, and there is no one else to ask... Thank you!

This is what the series looks like (attached)

The last temporary value is at the beginning of the file, the oldest one is at the end.

Do you think a data set would be enough? You need the method on which the data collection was based, what goals were set at the time of collection, etc.

[Maxim Dmitrievsky]

Evgeniy Chumakov:

Hello all!

If anyone has free time, please check the time series for predictability by neural networks.

As I do not understand anything in these ns, and there is no one else to ask... Thank you!

This is what the series looks like (attached)

The last temporary value at the beginning of the file, the oldest one at the end.

According to the five previous values:

>>> regr.score(X_train, y_train)
0.6002509612994398
>>> regr.score(X_test, y_test)
0.6010419196911408

60% accuracy

The 10 preceding values deteriorate, the accuracy is 50 percent.

on the 3rd one the accuracy is 57

Blue initial, orange prediction. Last 300 values

[Forester]

Maxim Dmitrievsky:

on five previous values:

60% accuracy

on 10 preceding deteriorates, accuracy 50

accuracy 57 on the 3rd.

Blue baseline, orange prediction. Last 300 values.

Something I don't understand. How can a regression model estimate in percentages? It should be in absolute values.

Here's the help I found https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LinearRegression.html :

score (X, y, sample_weight=None)[source]

Return the coefficient of determination R^2 of the prediction.

The coefficient R^2 is defined as (1 - u/v), where u is the residual sum of squares ((y_true - y_pred) ** 2).sum() and v is the total sum of squares ((y_true - y_true.mean()) ** 2).sum(). The best possible score is 1.0 and it can be negative (because the model can be arbitrarily worse). A constant model that always predicts the expected value of y, disregarding the input features, would get an R^2 score of 0.0.

I.e. it is not worth converting to percentages, because negative values are possible. Just a coefficient.

[Valeriy Yastremskiy]

elibrarius:
Something I don't understand. How can a regression model estimate in percentages? It has to be in absolute values.

logarithms

[Maxim Dmitrievsky]

elibrarius:

I don't understand something. How can a regression model estimate in percentages? It has to be in absolute values.

Here's the reference https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LinearRegression.html :

I.e., it shouldn't be converted to a percentage, because negative values are possible. Just a coefficient.

The negative ones are the inverse of the relationship. It's hard to imagine such a model

[iwelimorn]

Greetings all! Here is a simple animation system for changing parameters of your systems. Will be handy for debugging or observing the work of RL systems.

It consists of two scriptssave_csv.py and animation.py

All this works as follows.

Pre-paste the contents of the save_csv.py script into yourself. When called, the function will save your variables into a file with extension .csv.

Run the animation.py script

Then start your system, and the graphs will start.

The animation .py script can be left on while searching for a Gaal.

ripped from

https://www.youtube.com/watch?v=Ercd-Ip5PfQ

[Valeriy Yastremskiy]

Maxim Dmitrievsky:

negative is an inverse relationship. It's hard to imagine such a model

more is less than one) and you know how hard it is to explain a unit circle? More than one circle would have to be drawn... Although you're definitely not youche)))), but all is not lost))))

[Maxim Dmitrievsky]

Valeriy Yastremskiy:

more less than one) and do you know how hard it is to explain the unit circle? More than one circle would have to be drawn... Although you're definitely not a poke)))), but all is not lost))))

?