# Sir, how to calculate EER equal error rate for the following data

142 views (last 30 days)
Balaji M. Sontakke on 20 Sep 2018
respected sir, I have used following formula for calculating false acceptance rate FAR and FRR FAR=FP/(FP+TN) FRR=FN/(TP+FN) GAR=1-FRR my question is how to calculate equal error rate EER using following data
TP FP FN TN FAR FRR
Actual_class1 4 1 0 35 36 0
Actual_class2 4 4 0 32 33 0
Actual_class3 1 1 3 35 36 6
Actual_class4 4 1 0 35 36 0
Actual_class5 2 0 2 36 #DIV/0! 3
Actual_class6 4 0 0 36 #DIV/0! 0
Actual_class7 3 0 1 36 #DIV/0! 1.333333333
Actual_class8 2 1 2 35 36 3
Actual_class9 4 0 0 36 #DIV/0! 0
Actual_class10 4 0 0 36 #DIV/0! 0
mean 3.2 0.8 0.8 35.2
FAR 0.022222222
FRR 0.2
GAR 80 %

KALYAN ACHARJYA on 20 Sep 2018
Do you have any mathematical formula to calculate EER? if you have it, provide us, we can help you to implement in Matlab.
Balaji M. Sontakke on 20 Sep 2018
Sir, Is it possible to calculate EER using some formula in Matlab, how I can? please help me

KALYAN ACHARJYA on 20 Sep 2018
Edited: KALYAN ACHARJYA on 20 Sep 2018
Based on my basic knowledge: EER is the value where FRR and FAR intersect
So you can draw the plot FRR
Also, you can draw the plot FAR
The intersection of both curves gives the EER
Have you check here?
Also here

Balaji M. Sontakke on 20 Sep 2018
Sir, Firstly I want to know that, whichever formulas I used for calculating FAR and FRR are correct or not. secondly, I don't know the thresh values which I help us to calculate EER, so can you please guide me regarding it.
KALYAN ACHARJYA on 20 Sep 2018
Just plot two curves, there are multiple ways, you can find the intersection between two curves?
Any help needed, let me know here?
Balaji M. Sontakke on 20 Sep 2018
Sir, I plot the curve using the following code.. clc; clear all;
FAR=[0.027777778,0.111111111,0.027777778,0.027777778,0,0,0,0.027777778,0,0]; FRR=[0,0,6,0,3,0,1.333333333,3,0,0]; plot(FAR)
figure, plot(FRR)
idx = find(FAR - FRR < eps, 1); %// Index of coordinate in array px = FAR(idx); py = FRR(idx); figure, plot(px,py,'or','MarkerSize',18);