search

AU-BCE-EE / tric-fil-mod

A trickling filter model to simulate air treatment, written in Python
GNU General Public License v3.0
0 stars 0 forks source link

Visualizing model efficiency #30

Open AnneMortensen opened 4 months ago

AnneMortensen commented 4 months ago

Hoping to visualize model effeciency for multiple data sets and multiple model parameters (k and kl)

AnneMortensen commented 4 months ago

I think I have wrapped my head around what was meant by a scatterplot for this, and know how to make one. I think that would make sense to try first.

AnneMortensen commented 4 months ago

Made an attempt: https://github.com/AU-BCE-EE/tric-fil-mod/tree/7fdf1678e077082e48e46925c3992d1fd9bdf865/applications/Lab%20experiments/New%20structure/Tasks/6.8%20scatter%20plot

Do you see three pictures and the script I used to make them, @sashahafner ? And what do you think? The values for k and kl was - for my convenience - the ones found by a previous optimization round, so that is why they perform well (and better than the baseline of Onda and k=0). The k and kl values can be adjusted very easily, but the script takes about a cup of coffee to run pr set of new values tried.

sashahafner commented 4 months ago

I don't think those plots show too much, or show it clearly enough. At the moment I think this probably isn't a priority--it is good enough for you to just considering ME or MAE when selecting parameter sets to show and probably also listing the fit statistic values in the text for select parameter sets. But I'll show you one more idea for a plot.

sashahafner commented 4 months ago

image

But I don't think it should be a priority. Plotting and interpreting profiles, and sorting out analytical solutions is more important.