Numerical series density - page 19

New comment
 
First of all, understand for yourself the importance and relativity. What are we trying to find?
The cloud of highest density relative to the WHOLE ROW.

or in the whole row independent clusters
Which have only 1 single criterion, and that is proximity. There are no other criteria.
 
If you take a basin of water and put some cubes in it

From which cube the distance is shorter than all the others in total. That place will be heavier than any other point in the whole basin.
 
You have to understand, it's just a swing scale.

Bricks 1,3,5 on one side and 10,11,12 on the other

Where the density is greater per point is where 10,11,12

But which ends up outweighing all the bricks? 1,3,5. I haven't counted, but I think the centre of mass is 5.
 
What can we do.
- find the most significant cloud with the highest weight, relative to the whole row
-find independent clusters.
-find dependent clusters relative to the whole row.


What kind of example would you like me to give?
 
Now the question is different.

For myself, I identified the importance. One cluster. The centre of mass. I believe that this point will be the important level. Now the question.

What do we do with the same numbers? Ignore them? Or give that number some weight?


Ha, I know what to do with repeating numbers and how to give them weight. I'm waiting for your answers.
 
Vyacheslav Kornev:
You are either too clever or too clever.
I have found everything I can.
Again, formulate the problem, what is to be found.
Either the density of the number series,
Or a cluster,
Or the highest density cloud?

The density of the cluster (clump) in a numerical series. The number series itself is not interesting - it only acts as a definition of the limits.

Vyacheslav Kornev:
You have told that in the end we need the densest cluster, we have got it.

There is no certainty, as yet, that what we have got is the correct definition of the cluster....

Vyacheslav Kornev:
You understand that you have found exactly a cluster of ALL series of numbers.
If you want to find a set of clusters independent of the whole number series, you just use how close the numbers are to each other.

Precisely, I understand that you have found it and I've been telling you about it for a long time, but whether this information will be useful, is not clear yet.

Regarding "how close the numbers are to each other" - you need to automate the process of sifting out numbers that are not close - you need a criterion - logic. In my algorithm I sift out numbers by delta until they are less than half of the original series, but it can also be not enough - namely, the task - find the best criteria - one of the most difficult in this algorithm.

 
A dense set is a subset of space, the points of which can approximate any point of the enclosing space as well as possible.

So I'm right. The centre of mass and numbers around it within the most common delta is the densest set
 
Vyacheslav Kornev:
First of all, understand the importance and relativity. What are we trying to find?
The cloud of highest density relative to the WHOLE ROW.
or in the whole row independent clusters
Which have only 1 single criterion, and that is proximity. There are no other criteria.

You can't be so categorical - it hurts progress...

Vyacheslav Kornev:
If you take a basin of water and throw in cubes...
From which cube the distance is smaller than all the others in total. That place will be heavier than any other point in the whole basin.

It's clear - and I already wrote above, why it would be so... But, we need to consider the relation of each cube to the neighboring one...

VyacheslavKornev:
What can we do.
- to find the most significant cloud with the heaviest weight, relative to the whole row.
-find independent clusters.
-find dependent clusters relative to the whole row.
Which example would you like me to describe?

We can find it, but if we know what we're looking for... which makes it difficult to find it - the reason I gave above.

About the example - theoretically, let's take the numbers and:

1. Let us increase the number series by the same number series, previously multiplying it by 1000

2. the same point as 1, but replace 56 with 59

Vyacheslav Kornev:
Now the question is different.
For myself, I have determined the importance of. One cluster. The centre of mass. I believe that this point will be the important level. Now the question.
What do we do with the same numbers? Ignore them? Or give that number some weight?
Ha, I know what to do with repeating numbers and how to give them weight. I'm waiting for your answers.

I just add a minimum value of 1 point in my algorithm.

Files:
Test_FindOblast0.mq4  28 kb
 
I don't think it would be useful just to look for clusters. Better to calculate a landmark from the previous day, e.g. the centre of mass. That should suffice.
 
Vyacheslav Kornev:
Adense set is a subset of space, the points of which can approximate any point of the enclosing space as good as you want.

So I'm right. The centre of mass and the numbers around it within the most common delta is the densest set

I think I've already written about academic knowledge... Let's think within the limits of the task at hand, not theoretical inferences.

Clearly the number set is either one continuous integer or consists of regions, which are supposed to be ranked according to attributes, one of the attributes being density.

1...121314151617181920212223242526
New comment