I have with me currently a Cancer Registry dataset containing demographic and clinical data in 10 years. It has both continuous and discrete variables.
One of the things of interest is to find meaningful clusters within this dataset. These clusters are characterized by the different demographic data such as age, gender, ethnicity, occupation, etc. It is also part of the study to know which parameters strongly characterize these meaningful clusters as well. The clinician collaborator is interested to determine risk assessments and/or conduct predictive analysis on this dataset.
So, how can we implement the hierarchical clustering for this dataset?
.
there are quite a lot of approaches to (dis)similarity of mixed type data
one possibility (with R package) in the link below
http://stats.stackexchange.com/questions/15287/hierarchical-clustering-with-mixed-type-data-what-distance-similarity-to-use/15313#15313
now, as always with (dis)similarities, only you can judge if this makes sense in your context
.
Thanks for sharing. I guess this option is not easy to get, yet in the case that you somehow find a set of data that can be sent over, I offer to look it through for hidden patterns, key factors, indicators (here - biomarkers), multi-variable prediction and also early detection parameters as much as exist.
Good luck with that important work
Edith - Home of GT data mining
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA