Machine learning in trading: theory, models, practice and algo-trading - page 3378

New comment
 
Renat Akhtyamov #:

scrolling down to the balance.

I'm exactly the same way
 
Maxim Dmitrievsky #:
FF is nothing at all, nothing depends on it.
Most likely a terminological misunderstanding. But the rainbow reception is palpable.
 
fxsaber #:
Most likely a terminological misunderstanding. But the rainbow reception was perceptive.
or the question was asked incorrectly
Many people for some reason believe that they have come to a session to a psychic or psychologist :)
There is another category who piously believe that from the opinion is very important
 
Maxim Dmitrievsky #:
FF is nothing at all, nothing depends on it. It just matches inputs with desired outputs. I don't see what it has to do with robustness.

Fitness function - literal translation "health function", or fitness function, adaptability function. I.e. it is a way to evaluate some individual actions or the whole model. Also called evaluation function, it is applied either integrally or separately to individual elements of the system in the form of metrics.

In statistics: robustness (from robust - "strong", "strong", "firm", "stable") - a property of a statistical method that characterises the independence of influence of various kinds of outliers on the result of the study, resistance to interference. Robust method - a method aimed at identifying outliers, reducing their influence or excluding them from the sample.

As applied to MO and trading, robustness is the ability of a model (or TS) to demonstrate indicators on new data, indicators similar to those on the "training sample".

To assess the robustness of a model, various metrics, scores, influence functions are used, i.e. the fitness function is used.

In other words, the fitness function is a descriptive characterisation of what is required to be obtained. If the goal is to get a robust model, then we need to estimate robustness, compose a robustness fitness function. We need the maximum possible robustness (reliability, stability), so we need such a descriptive characteristic, fitness function, the maximum of which will correspond to the maximum robustness.


I hope it is now clear to everyone how the fitness function is related to robustness.

Examples of fitness functions:

Average of grades in a school. Can be applied as a normal average grade or weighted as in some countries.

Maximum allowable centrifugal acceleration as an estimate in the design of roads, including railways.

Next, as homework, you can continue the examples of fitness functions on your own, for understanding what it is and what it is used for.

 

Negative attitude to the notions of optimisation and fitness functions is due to the fact that the use of very simple fitness functions, such as balance, is widely practiced. This is an example of how the fitness function "final balance" chosen as an evaluation does not characterise the robustness of the TS in any way.

Fitness function is an evaluation method, nobody and nothing forbids to use any methods, not only those that were given by default by uncles.

 
fxsaber #:

The nature of the profit curve does not change by OOS: Size(OOS_Left) = Size(OOS_Right) = Size(Sample). In general, a result that cannot be passed by.

Stress tests are probably needed))))) Modelling black swans for TC)

 
Andrey Dik #:

Fitness function is a literal translation of "health function", or fitness function, adaptability function. It is a way to evaluate some individual actions or the whole model. It is also referred to as an evaluation function, applied either integrally or separately to individual elements of the system in the form of metrics.

In statistics: robustness (from robust - "strong", "strong", "firm", "stable") - a property of a statistical method that characterises the independence of influence of various kinds of outliers on the result of the study, resistance to interference. Robust method - a method aimed at identifying outliers, reducing their influence or excluding them from the sample.

In the case of MO and trading, robustness is the ability of a model (or TS) to demonstrate indicators on new data, indicators similar to those on the "training sample".

To assess the robustness of a model, various metrics, scores, influence functions are used, i.e. fitness function is used.

In other words, the fitness function is a descriptive characterisation of what is required to be obtained. If the goal is to get a robust model, then we need to estimate robustness, compose a robustness fitness function. We need the maximum possible robustness (reliability, stability), so we need such a descriptive characteristic, a fitness function, the maximum of which will correspond to the maximum robustness.


I hope now it is clear to everyone how the fitness function is related to robustness.

Examples of fitness functions:

Average of grades in a school. It can be applied as a normal average grade or weighted as in some countries.

Maximum allowable centrifugal acceleration as an estimate in the design of roads, including railways.

Next, as homework, you can continue the examples of fitness functions on your own, for understanding what it is and what it is used for.


Theorist! You've read the wrong books and you're littering the thread with off-topic stuff!

Here are examples of specific fitness functions:

ada(x, y,test.x,test.y=NULL, loss=c("exponential","logistic"),
                      type=c("discrete","real","gentle"),iter=50, nu=0.1, bag.frac=0.5,
                      model.coef=TRUE,bag.shift=FALSE,max.iter=20,delta=10^(-10),
                      verbose=FALSE,...,na.action=na.rpart)

Here's another one for NS

nnet(x, y, weights, size, Wts, mask,
     linout = FALSE, entropy = FALSE, softmax = FALSE,
     censored = FALSE, skip = FALSE, rang = 0.7, decay = 0,
     maxit = 100, Hess = FALSE, trace = TRUE, MaxNWts = 1000,
     abstol = 1.0 e-4, reltol = 1.0 e-8, ...)

Everything is written, you just have to learn how to use it. And if you learn how to use packages from R, then you will not be able to write such nonsense as " how fitness function is related to robustness" - robustness in MO is a separate and very serious problem.

Start using R and everything will fall into place and you will surprise the local public with examples of classification with less than 20% error and "robustness " outside the training file.

 
Loosely defined terms are evil)))))
 


 
СанСаныч Фоменко #:

Once again, the fitness function is a method of model estimation. The robustness of the model will depend on which fitness function is chosen for robustness estimation.

Robustness can also be viewed as an estimate of the model. Robustness is the evaluation of the model for its ability to perform on new data, i.e., it is the fitness function to be maximised.

I don't think these are very complex concepts that are difficult to comprehend, it is all the more surprising how many misconceptions there are on this topic.

Assessments, fitness function, can be tiered, with each tier controlling separate metrics. But for some reason, many people perceive the fitness function as something summarised, 'on top'.

1...337133723373337433753376337733783379338033813382338333843385...3512
New comment
Reason: