Cluster of customer

[andresfx_2000]

Hello friends,
I need your help, I tell you that I am creating a customer segmentation. and between the data comes fields of type string atributes , which with the algorithms that used it is not possible to include this type of data.
in python you can include it with k-means but in knime using the same algorithm I

can’t. What algorithm or node should I use?

cyty | id customer | game name | state
|Coquimbo|10073 |Baccarat |Vitacura|
|Coquimbo|10118 |Tragamonedas |La Florida|
|Coquimbo|10544 |Tragamonedas |Antofagasta|
|Coquimbo|10525 |Hold’em Poker |Maipú|
|Coquimbo|10329 |Tragamonedas |Rancagua|

[mauuuuu5]

Hi you can use three algorithms which are:

1. Expectation maximisation which is available in the Weka extension
2. The Kmodes which is available in klaR package in the R extension
3. The Kprototypes which is available in the clustMixType package in the R extension

I am attaching you a workflow that has both numerical and categorical information

Hope it helps

Mau

Kprototypes.knwf (3.2 MB)

[andresfx_2000]

thank you, very much i’m starting on this. I work for a casino in Chile. and I have a lot to learn

[gab1one]

Hi @andresfx_2000,
have you taken a look at the KNIME hub yet? There you can find example workflows for many applications, e.g. the following for customer segmentation (which is clustering customers):

https://hub.knime.com/knime/space/Examples/50_Applications/24_Customer_Segmentation_Use_Case/02_Customer_Segmentation_Use_Case

best,
Gabriel