I have the following experiment. My predictors are: Plant area, 6 types of plants, 5 time points and 12 replicates for class. Time points are not equally spaced. A normal plot of Area vs. time will show $6\times 12\times 5$ points in the plot which doesn't reflect the fact that 5 of those points correspond to the same plant at different time point but whose Area has changed with time. Is there a plot that shows some sort of derivative approach?
$\begingroup$
$\endgroup$
2
-
2$\begingroup$ How is PCA involved in the above description? $\endgroup$– chlCommented Apr 4, 2012 at 16:53
-
1$\begingroup$ I think the more appropriate terminology should be longitudinal data analysis (or repeated measures). It looks like you 72 cases with measurements repeated at five time points. I too think that you neglected to explain how you transformed the data to principal components. You haven't mentioned what the response variable is and the only continuous variable is plant area, the other two variables are discrete. Why would you apply principal components for this problem? It sounds like a repeated measures ANOVA problem. $\endgroup$– Michael R. ChernickCommented May 5, 2012 at 12:56
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
1
This seems like a good candidate for small multiples charting, where you have 6 charts (1 per plant type) and each of those plotted for area vs. time.
-
$\begingroup$ Or 6 different colors, with means line plotted, and error bars or confidence intervals. $\endgroup$ Commented May 14, 2012 at 2:38