K-Means Cluster Analysis in SPSS (SPSS Tutorial Video #30)

In this video I describe how to conduct and interpret the results of K-Means Cluster Analysis in SPSS. I especially emphasize using Hierarchical cluster analysis analysis to first determine the number of clusters in your data and then use that result as an input to the k-Means algorithm.

This SPSS tutorial series is designed to teach you the basics of how to analyze and interpret the results of data using SPSS. I will cover everything from the very basics of the main windows within SPSS, to manipulating data, to running and interpreting meaningful analyses like t-tests, ANOVA, regression, and many more, and visualizing results.

Follow me at:

Equipment Used for Filming:

[xeniavlasenko]

This is the 5th video I wath on K-Means and it FINALLY made sense. Thank you so much!

[bernardoluca]

Fantastic explanation! nothing to do with all those videos out there! keep going like this!

[nawilliam]

After a long search, finally something easy to understand

[jessicamartin]

Great! I was able to complete my entire assignment, using only this video

[GenuineReciprocity]

Your videos are so easy to understand and its so amazing how many people your kindness has been helping! I have a small question and was wondering if you can share your insight about it if you have time available. A study that I am trying to replicate has categorized individuals based on whether they score above or below the mean on two variables (i.e., high high, high low, low high, low low - 4 categories). I was advised that that technique was crude and that I should instead use a cluster analysis to categorize the groups. Why would cluster analysis be a better statistical analyses than what the original authors did in categorizing the variables? Sorry to trouble you! I look forward to more of your incredibly helpful videos!

[anass]

I really thank you for this great series of videos they have been so much useful in my research

[StevenWang]

Thank you very much, this video is very easy to understand !!

[transitionperf_MPO]

Thank you for a great explanation! I was wondering how to view demographic characteristics between each established cluster. For example, viewing percentage breakdowns of age, gender, etc. in each cluster. Thanks!

[tacs]

thank you so much! for this one and the hierarchical one!

[LXiao]

brilliant! thank you for uploading this video!

[ezeugochukukere]

This is very helpful. Oddly enough the reason i came across this video was because i was searching on how to calculate the initial cluster centers in SPSS.

I need them for my R script to perfectly replicate the K mean clusters analysis i run in spss...inputting the initial cluster centers calculated in SPSS provides the exact same results for the final cluster solution in R

...it was the first thing you said we don't need but i am pretty desperate in my search to find out how those initial cluster centers are calculated. Any help you could provide would be huge

[rypbzgw]

thank you for the great video, would you please explain about how to apply elbow method to find cluster number?

[katiesharp]

Hi I love your videos, really helping me analysis my dissertation data :) I was wondering if you had any videos that touched on how to identify the characteristics of your clusters? i.e. age, gender, those sort of things?

[aarinwood]

Great series of videos -- thank you! I do have one follow up question: What are the sample size requirements for Cluster Analysis? Thank you!

[zahraalinam]

Which method of hierarchical or K-means is the most appropriate for dichotomous variables with binary coding (0, 1) showing the presence and absence of a variable?

[vindaflyfox]

Hello, I am wanting to follow this process by doing a hierarchical cluster analysis to determine the k for my k-means analysis. My question is, my variables are not all on the same scale so in the hierarchical cluster analysis I will need to convert them into z-scores or something similar so they are comparable. How does this impact the k-means cluster analysis? Do I need to do an extra step here or will my variables already be converted and able to be used again after the hierachical analysis?

[zahraalinam]

In case the Sig for some variables is bigger than .001 what should we do? Should we screen and remove them and do the cluster analysis again?

[lingkan]

To cluster analysis for multimorbidity, is there any special format to arrange the data?

[divyajaiswal]

Can k means clustering data be represented graphically? If yes, how?

[tracyquetzal]

Hi Professor, very good presentation. I would like to know how can you understand your cluster in order to label them? What patterns do you find to classify your cluster?