Check for missing argument or incorrect argument data type in call to function 'fit' - using two 1x15 double arrays as an input argument

I am trying to fit to arrays and I get the error 'Check for missing argument or incorrect argument data type in call to function 'fit''.

The code I use:

clear all
close all
clc
 
year=[1991,1994,1995,1996,1997,2000,2001:1:2004,2007,2008,2016,2021,2024]
capacity=[ 0.4,0.5,0.5,0.7,0.6,2,2,2,3,4.5,5,5,10,12,14]
 
matrix=[year;capacity]
 
f=fit(year,capacity,'poly2')

Can anyone tell me what is wrong with this code?

[KSSV]

You can use polyfit instead.

year=[1991,1994,1995,1996,1997,2000,2001:1:2004,2007,2008,2016,2021,2024]
capacity=[ 0.4,0.5,0.5,0.7,0.6,2,2,2,3,4.5,5,5,10,12,14]
 

% f=fit(year,capacity,'poly2')
p = polyfit(year,capacity,2)

plot(year, capacity,'*r')
hold on
plot(year,polyval(p,year))

[Laura Stokbro]

Okay, Thank you for the answer!

The thing is that I would like to try also using an exponential fit to see what is more accurate.

I do not understand what is wrong with inputting two arrays into the 'fit' function?

[Laura Stokbro]

Thanks.

Though Ill have to predict future scenarios, so if the development is exponential it will grow faster than predictions using the polynomial fit. So I'd really like to be able to fit the points exponentially as well. Can you help me with that?

Answer 1

[Bernard Beitz]

Summary: Don't forget to install all needed toolboxes (curve fitting toolbox) after installing a newer version of matlab.

The Problem sounds like you have the statistics toolbox installed but not the curve fitting toolbox. Try typing

which fit -all

to validate. The command will show every function called 'fit' that matlab knows about The correct feedback is similar to this:

>> which fit -all
/MATLAB/toolbox/curvefit/curvefit/fit.m
/MATLAB/toolbox/stats/stats/@gmdistribution/fit.m  % gmdistribution method

If fit from curve fitting toolbox is missing, matlab will call the stats toolbox function which has different input types. If both are installed, matlab will determine by the type of the input paramters which function to call, so this won't be a problem.

Answer 2

[Alan Stevens]

Here's a way of fitting a simple exponential curve. However, you definitely don't want to use it to extrapolate!

year=[1991,1994,1995,1996,1997,2000,2001:1:2004,2007,2008,2016,2021,2024];
capacity=[ 0.4,0.5,0.5,0.7,0.6,2,2,2,3,4.5,5,5,10,12,14];
% Exponential fit by taking logs first
% capacity = a*exp(b*year)
% ln(capacity) = ln(a) + b*year

lcap = log(capacity);
c = polyfit(year,lcap,1);
a = exp(c(2)); b = c(1);
capfit = @(t) a*exp(b*t);
t = 1991:2024;
plot(year,capacity,'o',t,capfit(t))
grid
xlabel('year'),ylabel('capacity')
legend('data','exponential fit')

Answer 3

[JP]

I get the same error even with the examples from the Help Center: https://de.mathworks.com/help/curvefit/fit.html#bto2vuv-1-fitType

I tried for example this:

load census;

f=fit(cdate,pop,'poly1');

plot(f,cdate,pop)

Is there a problem with fit function in Matlab 2020a?

Best

[Ludovicamaria Amedeo]

you have to download the "curve fitting" toolbox :)

[JP]

thank you,

I luckily found it out by myself shortly after posting this. Strange that for such a basic method you need a toolbox...