Error Using fitnlm Function

function regresion_parametros_secador
clc, clear, clear all

TR  =[0.422103409 0.42981426 0.437540563 0.445266867 0.45299317 0.460719473 0.468445777 0.47617208 0.483898384 0.491624687 0.499350991 0.507077294 0.514803597 0.522529901 0.530256204 0.537982508 0.545708811 0.553435115 0.561161418 0.568887721 0.576614025 0.592066632 0.607519238 0.622971845 0.638424452 0.653877059 0.669329666 0.684782273 0.70023488 0.715687486 0.731140093 0.7465927 0.762045307 0.777497914 0.792950521 0.808403128 0.823855734 0.839308341 0.854760948 0.870213555 0.885666162 0.901118769 0.916571376 0.932023982 0.947476589 0.962929196 0.978381803 0.99383441 0.99993819];

HFG =[2500.9 2489.1 2477.2 2465.3 2453.5 2441.7 2429.8 2417.9 2406.0 2394.0 2382.0 2369.8 2357.6 2345.4 2333.0 2320.6 2308.0 2295.3 2282.5 2269.5 2256.4 2229.7 2202.1 2173.7 2144.2 2113.7 2081.9 2048.8 2014.2 1977.9 1939.7 1899.7 1857.3 1812.7 1765.4 1715.1 1661.6 1604.4 1543.0 1476.7 1404.6 1325.7 1238.4 1140.1 1027.3 892.7 719.8 443.8 0.0];

modelfun = @(p,HFG) (exp((p(1)+p(2)*(log(1./TR))^0.1+p(3)/TR.^2+p(4)/TR.^3+p(5)/TR.^4)^0.5));
beta0    = [0 53.63746882 0.01601773 0.031311254 0.071298912];
mdl      = fitnlm(TR,HFG,modelfun,beta0)

end

%ERROR MESSAGE%

Error using nlinfit (line 205)

Error evaluating model function

'@(p,HFG)(exp((p(1)+p(2)*(log(1./TR))^0.1+p(3)/TR.^2+p(4)/TR.^3+p(5)/TR.^4)^0.5))'.

Error in NonLinearModel/fitter (line 1123)

nlinfit(X,y,F,b0,opts,wtargs{:},errormodelargs{:});

Error in classreg.regr.FitObject/doFit (line 94)

model = fitter(model);

Error in NonLinearModel.fit (line 1430)

model = doFit(model);

Error in fitnlm (line 94)

model = NonLinearModel.fit(X,varargin{:});

Error in regresion_parametros_secador (line 14)

mdl = fitnlm(TR,HFG,modelfun,beta0)

Caused by:

Error using ^

One argument must be a square matrix and the other must be a scalar. Use POWER (.^) for elementwise

power.

>>

Answer 1

[Star Strider]

Use element-wise operations on every applicable operation in ‘modelfun’ and transpose ‘modelfun’ to return a column vector, and it works:

TR  =[0.422103409 0.42981426 0.437540563 0.445266867 0.45299317 0.460719473 0.468445777 0.47617208 0.483898384 0.491624687 0.499350991 0.507077294 0.514803597 0.522529901 0.530256204 0.537982508 0.545708811 0.553435115 0.561161418 0.568887721 0.576614025 0.592066632 0.607519238 0.622971845 0.638424452 0.653877059 0.669329666 0.684782273 0.70023488 0.715687486 0.731140093 0.7465927 0.762045307 0.777497914 0.792950521 0.808403128 0.823855734 0.839308341 0.854760948 0.870213555 0.885666162 0.901118769 0.916571376 0.932023982 0.947476589 0.962929196 0.978381803 0.99383441 0.99993819];
HFG =[2500.9 2489.1 2477.2 2465.3 2453.5 2441.7 2429.8 2417.9 2406.0 2394.0 2382.0 2369.8 2357.6 2345.4 2333.0 2320.6 2308.0 2295.3 2282.5 2269.5 2256.4 2229.7 2202.1 2173.7 2144.2 2113.7 2081.9 2048.8 2014.2 1977.9 1939.7 1899.7 1857.3 1812.7 1765.4 1715.1 1661.6 1604.4 1543.0 1476.7 1404.6 1325.7 1238.4 1140.1 1027.3 892.7 719.8 443.8 0.0];
modelfun = @(p,HFG) (exp(sqrt(p(1)+p(2).*(log(1./TR)).^0.1+p(3)./TR.^2+p(4)./TR.^3+p(5)./TR.^4))).';
beta0    = [0 53.63746882 0.01601773 0.031311254 0.071298912];
mdl      = fitnlm(TR,HFG,modelfun,beta0)

producing:

mdl = 


Nonlinear regression model:
    y ~ F(p,HFG)

Estimated Coefficients:
          Estimate      SE        tStat       pValue  
          ________    _______    _______    __________

    b1    -5.3764     0.53639    -10.023    6.2155e-13
    b2     80.604      1.2895      62.51    1.2803e-44
    b3    -12.018      1.5173    -7.9205    5.2486e-10
    b4     6.8052      1.0612      6.413    8.3294e-08
    b5    -1.1397      0.2096    -5.4375    2.2468e-06


Number of observations: 49, Error degrees of freedom: 44
Root Mean Squared Error: 12.4
R-Squared: 1,  Adjusted R-Squared 1
F-statistic vs. zero model: 2.6e+05, p-value = 3.84e-97

.