Water Cooler

[aj_954]

Hi,

I have 11,000 data points, lets call them x1 and y1.  I want to show case off in a scatter plot. Each of these points is labeled to a "cluster". The clusters are labeled 0-8.  About 10,600 of these poitns belong to cluster 1 and the rest belong to the remaining clusters about 10-20 points per cluster. When i make the dimesion "cluster" and the X measures x1 and Y measdures y1, I get " The chart is not displayed because it contains only undefined values. 

However, when I make the x and y dimensions the sum of the clusters, I get them to plot, however, it takes the sum of each x and y values per cluster, which is not what I want. I want to showcase all the data points with the color being from the cluster.

What am i doing wrong? 

[stevejoyce · Accepted solution]

Your "bubble" should be your data_point field (if you don't have an id/attribute for them, create an id field using rowno() in your data model), that will display a point for each data_point.

Your measures:  x-axis = sum(x1)

y-axis = sum(y1)

in appearence > colors and legent > 

switch Colors to custom and choose color by expression.

your expression can be:

pick(cluster + 1, rgb(color1), rgb(color2), red(), blue(), yellow(), rgb(color6))  ... etc.

[aj_954]

Hi, ive tried this. This produces a denisty plot. I am not looking for a denisty plot but a way to visually see all 11k points without having to zoom in. 

[stevejoyce]

If you want to view > 1000 points (1000 is the limit on scatter plot before it shows density) that's a different issue.

Natively the workarounds can be using a drill down dimension or enable Limitation to only show Top 1000 (ex. based on your first measure expression).

I'm not sure if there is a 3rd party extension to handle this.