Regression line on log-log plot with error bounds?

Hi everyone,

In my dataset, both x and y parametere have mean value, lower bound and upper bound. May someone suggest me how can i generate the plot attached below using my data set.

(1) Scatter plot on log-log scale with regression line

(2) upper and lower bound

....

clear all
clc


PDS=load('N_mean_bounds.txt'); % PDS loer mean and upper bound
pds_lb=PDS(:,1)/1000000; % lower bound
pds_m=PDS(:,2)/1000000; %mean value
pds_ub=PDS(:,3)/1000000; %upper bound


R=load('tes222t.txt'); %R lower bound, mean and upper bound 
r_lb=R(:,1); %lower bound
r_m=R(:,2); % mean value
r_ub=R(:,3); % upper bound

loglog(pds_mean,r_mean,'c','MarkerFaceColor',[0 0.447 0.741])

[Adam Danz]

Learn about confidence and prediction bounds here

https://www.mathworks.com/help/curvefit/confidence-and-prediction-bounds.html

You can either fit your raw data to a log function or fit the log transforms of your raw data to a line which is what I'd probably do.

[dpb]

While could make up data, be simplest if you would attach the data as a .mat file...

Looks to me like what was done was done incorrectly -- that set of error bands looks way to tight given the data unless they're to represent the error on the mean, not a prediction -- but even then, just can't be right. Probably not based on transforming the model to the linear space and computing but just the error on the log instead.

And, define which upper/lower bound you have in mind...