line builder with condition and Geometry cleaning

@jdh...i have tried it already with the Intersector...but it doesn't give me the result i want.

 

for example....

i have 4 lines class3

first line from 200 to 1800

 

second from 250 1850

 

third from 300 to 1900

 

fourth from 350 to 1950

 

 

so in this case i want to hold a line from 200 to 1950 with class 3 BUT:

there are lines with class 2

 

e.g.

 

from 400 to 2000

 

from 450 to 2050

 

from 500 to 3000 and so on

a Intersector gives me 50m segments

 

so class3 from 200 to 400.....

 

and from 400 to 1950 i get segments with class 2

and that is wrong

so Intersector is euther the false Transformer or i haven't chekced a param inside the transformer i can use for that

Greetz

Franco

 

 

 

franco69 wrote:

@jdh...i have tried it already with the Intersector...but it doesn't give me the result i want.

 

for example....

i have 4 lines class3

first line from 200 to 1800

 

second from 250 1850

 

third from 300 to 1900

 

fourth from 350 to 1950

 

 

so in this case i want to hold a line from 200 to 1950 with class 3 BUT:

there are lines with class 2

 

e.g.

 

from 400 to 2000

 

from 450 to 2050

 

from 500 to 3000 and so on

a Intersector gives me 50m segments

 

so class3 from 200 to 400.....

 

and from 400 to 1950 i get segments with class 2

and that is wrong

so Intersector is euther the false Transformer or i haven't chekced a param inside the transformer i can use for that

Greetz

Franco

 

 

 

The default setting on the Intersector for Attribute Accumulation Mode is Use Attributes From One Feature. That will give you the value of a random feature when there are multiple overlaps, which is not what you want.

 

 

That is why I specified you need to generate a list. Which will produce a list element for every overlapping feature. You can then sort the list to have the highest class be the first element.

jdh wrote:

If you send your lines through an Intersector, generating a list containing the class value. You can then sort the list based on the highest class (ListSorter) and promote the first value (ListIndexer, 0).

 

 

That should give you a line set with no overlaps, containing the highest class value. You can then send that to the LineCombiner (combine on attributes - class).

@jdh

after this the result is the right one...BUT: i don't exactly udnerstand why?

 

why i have 32 more features after the Intersector (exactly one segment long)?

what exactly happens in the ListSorter and ListIndexer Transformer then?

i want to undertsand this...

 

Greetz and Cheers

Franco

franco69 wrote:

@jdh

after this the result is the right one...BUT: i don't exactly udnerstand why?

 

why i have 32 more features after the Intersector (exactly one segment long)?

what exactly happens in the ListSorter and ListIndexer Transformer then?

i want to undertsand this...

 

Greetz and Cheers

Franco

The intersector will break apart features at intersections, including co-linear overlaps.

 

 

So if you had two non-parallel features that intersected each other at a single point, you would end up with 4 features.

 

 

If you had two features that partially overlapped one another (your case), you would end up with 3 features: the non-overlapping portion of each of the original features, as well as a single feature for the overlap.

 

 

By default, the overlapping segment will be assigned the attributes of one of the original features. Since we can't guarantee that's the 'higher class', we want to keep all the attributes from all the overlapping features so we can determine the higher class.

 

 

The generate list option on the intersector allows up to keep all the overlapping attributes.

 

 

Since you want the highest, the easiest way to analyse the list is to sort it, so that the elements on the list are ordered by the class value. That way the attributes from the feature with the highest class is guaranteed to be the first element in the list.

 

 

The listIndexer takes one element of the list (we want the first, so index 0) and 'promotes' it to regular attributes.

 

 

Original Data

ID: A

 

Class: 1

 

 

ID: B

 

Class:2

 

 

After Intersector

_list{0}.ID: A

 

_list{0}.Class: 1

_list{0}.ID: A

 

_list{0}.Class: 1

 

_list{1}.ID: B

 

_list{1}.Class: 2

 

 

_list{0}.ID: B

 

_list{0}.Class:2

 

 

After ListSorter

_list{0}.ID: A

 

_list{0}.Class: 1

_list{1}.ID: B

 

_list{1}.Class: 2

 

_list{0}.ID: A

 

_list{0}.Class: 1

 

 

_list{0}.ID: B

 

_list{0}.Class:2

 

 

After List Indexer

 

 

ID: A

 

Class: 1

 

 

ID: B

 

Class:2

 

 

ID: B

 

Class:2