Interpolation, approximation and the like (alglib package) - page 6
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Perhaps the most appropriate word would be "formula". On the one hand, there is a function defined by a data table, and on the other hand, there is a function defined by a formula.
I am watching your dialogue with interest. I would like to ask, what role do you assign to regression?
Perhaps the most appropriate word would be "formula". On the one hand, there is a function defined by a table of data, and on the other hand, there is a function defined by a formula.
The war of terminology is interesting as hell, but does anyone want to talk about the essence of the issue?
For example, this http://alglib.sources.ru/interpolation/inversedistanceweighting.php has already fallen away, because it's impossible to extract values from new data and get a f-u into which something can be fed. Roughly speaking, it does not work on new data.
You don't seem to have practically solved the problem of interpolation, do you? Yes? In interpolation, you don't talk about simplifying a function. The point of interpolation is not to simplify. Someone lumped interpolation and approximation under one heading in a textbook and there you go...
Why specify the domain of a function that is already defined from minus infinity to plus infinity?
As mathematical gurus write textbooks - a dump of all in one chapter, so on these textbooks senior lecturers give lectures and the same dump goes to the heads of students, some of which later become teachers and the cycle closes. Then some of them, instead of explaining the meaning of established definitions, introduce new ones... instead of a function, it's a mapping and it's a dead end. Some are loaded with all this terminology and think they have become mathematicians... some kind of disease of leftism in communism.
When we are dealing with Maksim, you have to understand that for all his literacy he lacks accuracy of thought and precision in terminology definition. Because of this it is completely impossible to understand what he wants.
Here is an example of this thread.
Maxim is looking for spline parameters. He defines these parameters by a set of points.
What is this? The answer is obvious: the man is engaged in approximation and this statement is approximate since it is not clear from his posts what he has as input: a function defined in a table or a set of random values for which he is looking for analytical form with an error. Since we and Maxim deal with quotes, it is most likely an approximation, i.e. values of initial points can only be given by the function with an error.
The situation is aggravated because the very ideology of splines requires specification of inflection points where pieces of this very spline will be joined in a clever manner.
So what are we talking about? What points are we talking about?
Next. How will this function be used in analytical form? It is not at all clear from the posts, because the resulting function can be used to obtain values INTERFLOWING values from the table or WITHIN them. In the first case it is interpolation, in the second case extrapolation, which is very attractive to us because extrapolation is something else than prediction.
Maxim has made a 5-page mess with his inaccurate problem statement, when it is so simple.
When we are dealing with Maxim, you have to realise that for all his literacy, he lacks precision in his thinking and accurate definition of terminology. This makes it completely impossible to understand what he wants.
Here is an example of this thread.
Maxim is looking for spline parameters. He defines these parameters by a set of points.
What is this? The answer is obvious: the man is engaged in approximation and this statement is approximate since it is not clear from his posts what he has as input: a function defined in a table or a set of random values for which he is looking for analytical form with an error. Since we and Maxim deal with quotes, it is most likely an approximation, i.e. values of initial points can only be given by the function with an error.
The situation is aggravated because the very ideology of splines requires specification of inflection points where pieces of this very spline will be joined in a clever manner.
So what are we talking about? What points are we talking about?
Next. How will this function be used in analytical form? It is not at all clear from the posts, because the resulting function can be used to obtain values INTERFLOWING values from the table or WITHIN them. In the first case it is interpolation, while in the second case it is extrapolation, which is very attractive to us because extrapolation is something else than prediction.
Maxim made a mess of his inaccurate formulation of the problem for 5 pages, while everything is so simple.
Given: a set of features for a non-grid. We need to transform them in many different ways. I don't care what name it will have, as long as I have plenty of transformation variants. I pick the best one out of them and save it as a function/formula/whatever. Then I need to insert raw data points into it (1 or more) and get transformed value. This point can either lie inside the set where the original conversion was done, or it can be an outlier
There are 2 approaches: transform each feature individually, or all at once in a bundle
There are kernel transformations of features via polynomials, I don't know how to do them myself.The terminology war is of course so fucking interesting, but is anyone willing to speak on the matter?
...
In order to get to the heart of the matter, you must first understand the question. If the task is to interpolate, there is no arbitrary choice of the number of node points.
I am watching your dialogue with interest. I would like to ask, what role do you see regression playing?
Why should it play a role? Basically, it is an approximation.
In order to get to the heart of the question, you must first understand the question. If the problem is interpolation, there is no arbitrary choice of the number of node points.
I wrote above in response to Sanych
Given: a set of features for a non-network. We need to transform them in many different ways. I don't care what it will be called, as long as there are many variants of transformations. Then we need to have some f-function, in which the raw points (1 thing) get its new values. This point can either lie inside the set where the original transformation was done, or be an outlier.
How many in numbers? How many features? How many inputs does the neural network have?
Is "1 thing" a function, or a single point thing?
How many in numbers? How many signs? How many inputs does the neural network have?
Is "1 thing" a function, or a single point thing?
Any number of inputs.
You can do a separate converter for each one separately or a generic one that changes them together at once
1 item is one point
Given: a set of traits for a non-network. We need to transform them in many different ways. I don't care how it will be called, as long as there are many variants of transformations. Then we need to have some f-function, in which the raw points (1 thing) get its new values. This point can either lie inside the set where the original transformation was done, or be an outlier.
My URMhttps://www.mql5.com/ru/articles/250 is the best way to do this, although I'm not a neural network supporter.