# seek help on clustering analysis

I have about 79,000 game players' data and we are trying to cluster these players into different classes. But so far we did not get a consistent cluster solution (we used K-means clustering). I plotted the data with two almost uncorrelated important variables identified by PCA, shown below. I also tried log transformation of the variables, which did not help much either. any suggestions? Maybe it is impossible to have a little overlapped clustering.

I used SAS to run the cluster analysis, and the ccc/pseudo F/Pseudo T^2 indicate there are 4 clusters:

but most of the players were not included in these clusters (because the "Distance Between Centroids" is 0 for them ):

So, does this mean there is no way to separate them into different groups?

• We can't do much with just a plot: at a minimum, please describe the data and explain what you mean by a "consistent" clustering. – whuber May 24 '16 at 22:05
• The data is gaming play data in a whole year. we have 79000 players (rows), with variables like how much money spent, how much time spent, how many times they played, how many days they played in that year, average money each play, average played times each day, etc. By saying not consistent, I mean using different combination of the variables, we got 3 clusters, 4 clusters, 5 clusters, 6 clusters, etc. Some variables are highly correlated, so tried combinations of less correlated variables. – ziweiguan May 25 '16 at 1:21
• Could you explain the purpose of your analysis? Since you don't seem to have definite variables in mind, just what are you hoping the clusters will tell you? – whuber May 25 '16 at 3:04
• we want to classify the players into different groups - the main goal is to identify those high-risk game player -- their betting pattern etc – ziweiguan May 25 '16 at 3:52
• Since there's nothing in your question that provides any information that seems relevant to that objective, it's difficult to see how it could be answered. – whuber May 25 '16 at 3:57