Optimising a function with multiple inputs

vasile stratan

2020 10 10

0 个回答

2 次查看（30 天）

0 个投票

Hi everyone.

I have written a function to detect outliers in data sets based on tresholds defined by relative and absolute tolerances (tol_tr, tol_ta)

[outlier_detected]= find_out(outlied_n, TLE1.Frac_year, gap_start1, gap_end1,sample,tol_tr,tol_ta);
f_n = length(setdiff(x_out,out_detected))/nr_out;                                % normalized quantity of false negatives 
f_p =length(setdiff(out_detected,x_out))/(length(outlied_n)-nr_out);             % normalized quantity of false positives 
rad=sqrt(f_n^2+f_p^2);                                                           % distance from origin

An outlier is detected if two quantities inside the find_out function are above the tol_tr and tol_ta. f_n is the quantity of missed outlier and f_p is the quantity of false outliers.I want to find the best tol_tr and tol_ta, to minimise the f_n and f_p. The best solution is the one closer to origin, so rad is expected to be minimized. Would be thankful for some help