Sep 27
I changed the outputs so that it gives a data frame with the variable being changed and the return variable, but I'm having some trouble on thinking about how to approach the quadrat optimization best.
I've been looking at variance/population like we talked about rather than just variance, which is generally giving more sensible results. I'm also thinking about changing the sample size to a general cost function, which would be (cost of sampling a quadrat/cost of sampling an individual)*quadrats sampled+sample size. This idea is taken from the appendix of the optimization paper that I posted on the site, and accounts for the work needed to go between quadrats.
I've also been thinking about finding the weight for variance and sample size (cost function). We talked about doing this analytically, but I'm not really sure how to think about that.
We had also talked about what exactly is being changed in the optimization. Originally I was thinking total number of quadrats in the population and the percent sampled, but because population characteristics (size and clustering, specifically) affect the optimal combination of these we talked about using different parameters, but I'm not sure what the best parameters would be.
