Good evening,
This is my first time using the Dimension Reduction tool and unfortunately, there really aren't any Esri videos that talk about this tool. So I have census tract level data containing 30 different variables (covering race, age/sex, income, housing, health, total population, education) and I was trying to reduce these variables by splitting them into 7 components using the Dimension reduction tool. I ran the tool with a number of permuations of 99 and the tool ran fine, however I'm finding it difficult to delineate which variabales belong to which component. With that, these are my questions:
1. Using the eigenvalues table and chart (I have attached them), how can I tell which variables make up each component? The goal is to use these 7 components for a flood vulnerability analysis (I want to use an ANOVA to see if there any differences across these components within and outside the flood zones).
2. My understaning of the Broken stick value from this documentation (https://pro.arcgis.com/en/pro-app/latest/tool-reference/spatial-statistics/how-dimension-reduction-w...) is that, it tells me the optimal number of components. Based on my scree plot (also attached), the intersection was at '8' with a 70% total variance explained but in the output message window (also attached), a value of 3 was given. However when I ran the tool without inputting any number of components, that value changed to '28' with a 100% variance explained. Can some light be shed on this please as it has me very confused.
3. Lastly, for the Bartlett's test of Sphericity, the number of components is 28 (see attached output message window). Does this mean that only 28 variables are relevant?
Please any sort of assistance on this will be very useful and I will be extremely grateful.
Also, if you feel there is a better way to go about this in ArcPro, I'm more than open to working with your suggestions.
Thank you.
