Cluster analysis plot
Hello everyone,
I'm new on this forum but I use JASP a lot, both for research and teaching so I think I'll like it here !
I'm using cluster analysis (K-means) for the first time on two variables. I use JASP to draw the plot, and the output is this :
Now, one of my colleague says that "you need to make the dimensions apparent".
Is this true ? Can I and should I find out which dimension is on Y and X and make them appear (for a scientific paper) ? I somehow thought that because of the unsupervised nature of K-means clustering I'm not allowed to use labels. But if there are only 2 variables, I guess you can somehow them on a 2D graph.
I'm looking for your help on this question, and notably how to perform it on JASP if I have to make the label appear. Thank you very much and have a great day !
Comments
I'll forward this issue to our ML experts! Sorry for the tardy response.
Cheers,
E.J.
Hi Manhervart,
> I somehow thought that because of the unsupervised nature of K-means clustering I'm not allowed to use labels.
In general, that's true. A t-SNE plot is a high-dimensional projection of all the variables to make the results somewhat visually interpretable, so the axes don't have clearly defined interpretations or labels. But if you have only 2 predictors, you might as well make a scatter plot and label the points by their predicted cluster. You can do this in the following way.
1. Add the predicted clusters to the data set as a new variable, in my case "predictedClusters".
2. Open Descriptives and add your predictors under "Variables" and the predicted clusters under "Split".
3. Select "Scatter Plots".
4. You should now see a figure like this:
There are some options to customize the scatter plot (I removed the regression line for example). Also note that since the t-SNE is a projection, it may warp the raw data to some extent.
Hope that helps, if you have any more questions, let me know!
Don
Hi,
Thank you very much for this clear answer, and sorry for the late answer.
It definitely helps !
Have a great day
Manhervart