Hi Degan, I have a question, maybe you can help me! I'm doing same cluster analysis in enterprise miner, but I'm not sure what algorithm analysis use, there is no choice?.
Hi Degan In my first attempt after having filtered appropriately when I ran the clusters with proper settings, I got 10 Clusters. any reasons why I didnt get 4 as you. Please answer. Dev
The short answer is that if you used the same data as I did, and the default settings, you should have gotten 4 initially. If you changed any settings, such as specifying the number of clusters you wanted, then it would be forced to 10. Longer answer: By default, it starts at 50 clusters and then calculates a number (the CCC), and then consolidates clusters over and over again and it picks one of the peak values for the CCC. Assuming you had the data set that I did, it had a local maximum (View-Summary Statistics-CCC Plot) at 4, which is not necessarily a good result. So to get 10 you have to tell it to use 10 and go to the Properties panel of the Cluster node, select "Specification Method -> User Specify" and then set the maximum number to 10. It is important to consider that the number of clusters you want may be subjective, however, and so there isn't necessarily a wrong number to pick.
![เป็นไปได้ไหม - WanMai [Official MV]](http://i.ytimg.com/vi/_6lZ6b3_7Vo/mqdefault.jpg)
Hi Degan.
I found this piece of gem today. Thank you for delivering something my useless university teacher is paid for and fails.
thanks a lot!!! you helped me when i am struggling on my uni data analysis assignment.
Love your videos Degan. Keep posting please
It's very clear and helpful. Thank you!
Great job. Thank you for creating this!
Great video, thanks.
Hi Degan,
I have a question, maybe you can help me!
I'm doing same cluster analysis in enterprise miner, but I'm not sure what algorithm analysis use, there is no choice?.
Degan,
I liked this example. Where could I find this sample data set?
Found the folder and file in Studio.
How can you select the cluster methods? How can you look at R squares?
I’m just wondering why you excluded LocX & LocY?
I have the same question? I dont why he edited those specific variables
Hi Degan
In my first attempt after having filtered appropriately when I ran the clusters with proper settings, I got 10 Clusters. any reasons why I didnt get 4 as you.
Please answer.
Dev
The short answer is that if you used the same data as I did, and the default settings, you should have gotten 4 initially. If you changed any settings, such as specifying the number of clusters you wanted, then it would be forced to 10.
Longer answer:
By default, it starts at 50 clusters and then calculates a number (the CCC), and then consolidates clusters over and over again and it picks one of the peak values for the CCC. Assuming you had the data set that I did, it had a local maximum (View-Summary Statistics-CCC Plot) at 4, which is not necessarily a good result. So to get 10 you have to tell it to use 10 and go to the Properties panel of the Cluster node, select "Specification Method -> User Specify" and then set the maximum number to 10. It is important to consider that the number of clusters you want may be subjective, however, and so there isn't necessarily a wrong number to pick.
Degan Kettles Thank you. This is helpful. Much Appreciated. Dev