Implement hierarchy clustering

CIAT-DAPA / subsets_genebank_accessions

Web tool which allows creating subsets of accessions based on indicators

0 stars 1 forks source link

Implement hierarchy clustering #58

Closed stevensotelo closed 3 years ago

stevensotelo commented 3 years ago

@khadijaAouzal

[x] Develop a method to build a method

stevensotelo commented 3 years ago

@khadijaAouzal

She is exploring information about how we can do
[x] She will compare the two algorithms HDBScan and agglomerative

@bmora-0110

He checked the DBScan Hierarchy

stevensotelo commented 3 years ago

@khadijaAouzal

She compared the two algorithms HDBScan and agglomerative
[x] She will send output data to @bmora-0110
[x] She develop a function which include HDBSCan and agglomerative. The function

@bmora-0110

[x] He will check the outputs of the algorithms and give us his thoughts

stevensotelo commented 3 years ago

@khadijaAouzal

She sent the data to @bmora-0110
She develop the function
[x] She is debugging the function and then she will upload the function to github

@bmora-0110

He checked the data.
[x] He will write the questions here

stevensotelo commented 3 years ago

@bmora-0110

[x] He will write the questions here

@khadijaAouzal

She developed the function and uploaded the script

bmora-0110 commented 3 years ago

Kjadija, I have some comments here:

1 . We are going to implement two methodologies DBSCAN and HDBSCAN, I understand that the methodologies are different but, I consider that the results should be more or less similar. So I think it would be important to do a test if this is really true.

There is a term in the algorithms called "Border points" and in our case they are accessions that will not belong to any cluster, I think this would be fine for DBSCAN, but for HDBSCAN, I consider that it is of great importance that all the accessions are included in the different clusters.

Thanks.

stevensotelo commented 3 years ago

We continue with the discussion and we will wait for the comments from users to define hyper parameters

stevensotelo commented 3 years ago

@khadijaAouzal

She developed the function