Linear regression in x log plot matlab

Hey,

I would like to show the linear regression (straight line) of this data with the x-axis in log . So far I only get a mossy line.

Can I ask for a little help?

y = [156.547 173.256 126.514 131.344 158.640 105.313 132.794 134.500 130.340 126.969 91.931	107	104	75.717 85 65.500 145];

x = [36000 21000 18300 61300 53000 65000 37000 160000 470000 43000 93000 160000	82000 214000 120000	160000 40000]

x = 1×17

36000 21000 18300 61300 53000 65000 37000 160000 470000 43000 93000 160000 82000 214000 120000 160000 40000

mdl = fitlm(x,y);

hold on
plot(mdl)
set(gca, 'XScale', 'log')

Answer 1

[Praveen Reddy]

Hi Reto,

I understand that you intended to display the linear regression of the data with the x-axis in a logarithmic scale. However, you encountered a situation where the resulting line appeared curved instead of being straight. The reason you are seeing a curved line instead of straight line in your regression plot is because you are using a logarithmic scale for the x-axis. When you apply a logarithmic scale to the x-axis, the relationship between the x-values and the y-values will no longer be linear. To obtain a straight line in your regression plot, you should fit the model using a logarithmic transformation of the x-values. Please refer to the following modified code:

y = [156.547 173.256 126.514 131.344 158.640 105.313 132.794 134.500 130.340 126.969 91.931 107 104 75.717 85 65.500 145]; 
x = [36000 21000 18300 61300 53000 65000 37000 160000 470000 43000 93000 160000 82000 214000 120000 160000 40000]; 
mdl = fitlm(log(x), y);  % Fit linear regression with logarithmic x-axis 
plot(mdl);  % Plot the linear regression line 
set(gca, 'XScale', 'log');  % Set x-axis to logarithmic scale 
ylim([20, max(y)]);  % Set the y-axis limits from 20 to the maximum value of y 

Modifying “mdl=fitlm(x,y)” to “mdl=fitlm(log(x),y)” will ensure that the relationship between the “x-values” and the “y-values” is linear, resulting in a straight line in the plot.

Hope this helps!