lundi 17 décembre 2018

Which method to use? Find patterns or clusters within variables

I have some data like this, for five different locations. Concentrations of various compounds and discharge. This data is from location 1. The real data contains more compounds and sample dates.

Date          Compound A    Compound B    Compound C     Q
01-01-2015       0.1          1             0.02         200
01-06-2015       0.3          0.4           0.05         250
01-12-2015       0.3          1             0.1          500
01-01-2016       0.5          0.7           0.05         300
01-06-2016       1.2          0.3           0.4          50
01-12-2016       0.8          0.8           0.06         70
01-01-2017       0.5          1.3           0.04         250
01-06-2017       0.8          0.5           0.2          150
01-12-2017       0.5          0.9           0.1          100

I want to find the relation between the compounds based on the variable Q. For example, see if various compounds behave the same way when Q is below 200. Find patterns within the compounds based on Q.
I would like to cluster the compounds somehow, first based on Q and then add the variable of location and relate them.

I'm not sure which method to use. I was thinking of self organizing maps, but I think I need more variables. I would like to do this first for one location. Do you know any method that could help me find patterns or clusters based on other variable?

Thank you!

Aucun commentaire:

Enregistrer un commentaire