Q: How should I present results if I have multiple variables?

Detailed Question -

I have worked on a forecasting technique. I have used 3 gauging stations and each station has 3 parameters and 3 seasons. So in total I have forecasted the data for 27 variables. Now I feel that 27 plots are not possible to include in a manuscript. So what can be done?

1 Answer to this question

From your question, I'm not really sure if you have 27 variables. If there were 3 parameters tested in all 3 stations, then I feel there should be 3 variables only, not 27.  Also, I'm not sure what you mean by plots in this case. 


Plots typically suggest scatter plots, which are used to represent correlation. Do you want to know if you can add 27 scatterplots? This is slightly odd, as scatterplots typically indicate relation between two variables (bivariate scatterplots). However, scatterplots with 2 X axes or 2 Y axes may also be plotted, which would involve 3 variables.


If you mean that you want to plot (i.e., display correlation among) station x parameter x season, you can use 3D scatterplot. This is a multivariate scatterplot for 3 variables each with continuous data.  


If you actually have more than 3 variables, you could go for a scatterplot matrix, in which relation between each pair of variables is plotted in a big matrix like layout. Check this example


Alternately, if you have 27 variables, Principal Component Analysis (PCA) would also help.


Related reading: