Using Curve Fitter, computing Local Maximum (y) and corresponding (x) and export x y of fitted curve to external format (Excel)

I have a dataset: matlab_test_data.xlsx which I have uploaded to MATLAB (Home > Upload > matlab_test_data.xlsx).

I plotted the data using the following command:

%: Read the excel file using readtable function
rawTable = readtable('matlab_test_data.xlsx','Sheet','Sheet1');
x = rawTable.Days; %: get the excel column, Header1 (header name)
y = rawTable.Sucrose; %: get the excel column, Header2 (header name)
figure;
plot(x,y);

I then used the Curve Fitter (Apps > Curve Fitter > Select Data > Select Fitting Data)

Fitname: Trendline

X data: x

Y data: y

(Ignoring NaNs in data)

Polynomial Degree: 3

Robust: Off

Followed by Export > Export to Figure and Export to Workspace

How can I find the Local Maximum (y): 12.0799 and corresponding (x): 228.885 from the Fitted Curve (Trendline) in Figure 2 and extract (x,y) values of the Fitted Curve to Excel.

Answer 1

[Matt J]

If all you're after is the local max, I would skip the curve fitting app and use polyfit.

keep=~(isnan(x)|isnan(y));
x=x(keep);
y=y(keep);

p=polyfit(x,y,3);
dp=polyder(p);
ddp=polyder(dp);

xs=roots(dp);
xs(abs(imag(xs))>1e-6)=0; %get rid of imaginary junk

xmax=xs(polyval(ddp,xs)<0)

xmax = 229.1196

ymax=polyval(p,xmax)

ymax = 12.0800

writematrix([xmax,ymax],'filename.xlsx')

[Hansraj]

Thank you.

As an alternative to Apps > Curve Fitter, I used the following to plot the trendline:

include=~isnan(y);
fobj = fit(x(include),y(include),'poly3')

The resulting output is as follows:

include=~isnan(y);
fobj = fit(x(include),y(include),'poly3')

fobj = 

     Linear model Poly3:
     fobj(x) = p1*x^3 + p2*x^2 + p3*x + p4
     Coefficients (with 95% confidence bounds):
       p1 =  -2.367e-05  (-4.589e-05, -1.448e-06)
       p2 =      0.0145  (0.0009933, 0.02801)
       p3 =      -2.917  (-5.614, -0.2204)
       p4 =       203.9  (27.05, 380.8)

Followed by

       plot(fobj,x,y)

How can I export the (x,y) data of plotted trendline to an Excel format.

[Matt J]

What about your original quesion about finding the local max, which I posted an answer for?

[Hansraj]

Thank you very much.

I used the provided solution and obtained targeted result. I notice you have edited the command. Is there any particular reason.

To obtain the (x,y) data of the line, I have selected the line and used the following command:

x=get(gco,'Xdata')
y=get(gco,'Ydata')

The results are as below. Is there a command to export the (x,y) data to excel format.

Matt J Edited

[Matt J]

I used the provided solution and obtained targeted result.

I'm glad, but please Accept-click the answer if that's the case.

I notice you have edited the command. Is there any particular reason.

I made many iterations, so I'm not sure which specific change you are referring to. One change I made was to use the second derivative of the polynomial to distinguish the max from the min.

The results are as below.

I've abbreviated it to make the post more readable. However, be mindful tha you can evaluate the fit at any y values you wish

yfit=fobj(x);

Is there a command to export the (x,y) data to excel format.

Yes, you can use xlswright or writematrix.

[Hansraj]

I'm having trouble with the below command and recommended xlwrite and writematrix to export the (x,y) data to excel format.

x=get(gco,'Xdata')
y=get(gco,'Ydata')

[Hansraj]

To those who may be interested, please read

x=get(gco,'Xdata')
y=get(gco,'Ydata')

Followed by:

writematrix([x(:), y(:)],'filename3.xlsx')