how to convert this code exposal to linear and how to find mape

function [a,g0] = getExpontialModel(f)
A = zeros(1,length(f)-1);
G_0 = zeros(1,length(f)-1);
for i=2:length(f)
    if f(i-1) ~= 0 
        A(i-1) = f(i)/f(i-1);
    else
        A(i-1) = 1;
    end
end
a = median(A); %more robust than mean function
for i=1:length(f)-1
    G_0(i) = f(i)/a^i;
end
g0 = median(G_0);%more robust than mean function

Answer 1

[Rahul]

Hi @andrew moisidi,

In order to convert the exponential model to a linear model, consider using the logarithmic transformation. For this 'log' function can directly be used.

In order to fins the MAPE (Mean Absolute Percentage Error), a direct function is not availble in MATLAB in R2017b, however, it can be calculated using its formula. Here is an example:

g0 = median(G_0);

% Linearize the data
log_f = log(f);
    
% Calculate predicted values using the exponential model
predicted_f = g0 * a.^(0:length(f)-1);

% Calculate MAPE
mape = 100 * mean(abs((f - predicted_f) ./ f));

From MATLAB R2022b a direct 'mape' function was introduced:

mape_1 = mape(predicted_f,f);

The following MATLAb Answers can be referred:

https://www.mathworks.com/matlabcentral/answers/844265-how-to-calculate-mape

https://www.mathworks.com/matlabcentral/answers/444360-calculation-mape-mean-absolute-percentage-except-for-some-values

The following MathWorks documentations can be referred to know more:

'log': https://www-test1.mathworks.com/help/releases/R2017b/matlab/ref/log.html

'mape': https://www.mathworks.com/help/matlab/ref/mape.html