estimating parameters by lsqcurvfit

Good afternoon sir

sir please tell me how to draw graph for infected data vs predicted for this lscurvefit code.

i wrote code like this

plot(time,infected ,time,Cv) but it shows error sir.

code is

data = [  1  25
    2  75
    3  227
    4  296
    5  258
    6  236
    7  192
    8  126
    9  71
    10  28
    11  11
    12  7];
time = data(1:12,1);
infected= data(1:12,2);
beta0 = 1;
lamdaa0= 0.019;
lamdas0= 0.0715;
etas0 = 0.03;
etaq0 = 0.04;
gammaa0 = 0.2;
gammaq0 = 0.13;
gammah0 = 0.07;
mua0 = 0.0001;
muh0 = 0.0002;
lb =[0,0,0,0,0,0,0,0,0,0]; ub = [1,1,1,1,1,1,1,1,1,1];
B0 = [beta0; lamdaa0; lamdas0;  etas0; etaq0; gammaa0; gammaq0; gammah0; mua0; muh0 ];
options=optimset('MaxFunEvals', 10000, 'MaxIter', 10000, 'TolFun', 0.0001, 'TolX',0.0001,'Display','on');
[B,resnorm,RESIDUAL,exitflag,OUTPUT,LAMBDA,Jacobian] = lsqcurvefit(@diff1,B0,time,infected,lb,ub,options);
disp(B)
%plot(time,infected,time,C)
function C = diff1(B,time)
x0 =  [1217378052,120,3,2,1,1,0,0];
[t,Cv] = ode45(@DifEq, time, x0);
    function dC = DifEq(t, x)
        N = 1390000000;
        pi = 700 ;
        zetaa = 0.1;
        zetas = 0.2;
        zetaq = 0.3;
        zetah = 0.3;
        omega = 0.2;
        theta = 0.5;
        mu = 0.0000425;
        beta = B(1);
        lamdaa= B(2);
        lamdas = B(3);
        etas = B(4);
        etaq = B(5);
        gammaa = B(6);
        gammaq = B(7);
        gammah = B(8);
        mua = B(9);
        muh = B(10);
        xdot = zeros(7,1);
        
        xdot(1) = pi -beta*(zetaa*x(3) +zetas*x(4) +zetaq*x(5)+zetah*x(6))*(x(1)/N) -mu*x(1);
        xdot(2) = beta*(zetaa*x(3) +zetas*x(4) +zetaq*x(5)+zetah*x(6))*(x(1)/N) -(omega+mu)*x(2); 
        xdot(3) = theta*omega*x(2)-(lamdaa+gammaa+mua+mu)*x(3);
        xdot(4) = (1-theta)*omega*E-(lamdas+etas+mu)*x(4);
        xdot(5) =  lamdaa*Ia+lamdas*Is-(etaq+gammaq+mu)*x(5);
        xdot(6) =   etas*Is+etaq*Q- (gammah+muh+mu)*x(6);
        xdot(7) =  gammaa*x(4) + gammaq*x(5) + gammah*x(6)
        
    end
C = Cv(:,2);
plot(time,infected ,time,Cv)
end

Error is

Unrecognized function or variable 'infected'.

Error in lscurvfit6>diff1 (line 72)

plot(time,infected ,time,Cv)

Error in lsqcurvefit (line 225)

initVals.F = feval(funfcn_x_xdata{3},xCurrent,XDATA,varargin{:});

Error in lscurvfit6 (line 19)

[B,resnorm,RESIDUAL,exitflag,OUTPUT,LAMBDA,Jacobian] = lsqcurvefit(@diff1,B0,time,infected,lb,ub,options);

Caused by:

Failure in initial objective function evaluation. LSQCURVEFIT cannot continue.

Answer 1

[Torsten]

Although this has been answered several times already, here we go again:

data = [  1  25
    2  75
    3  227
    4  296
    5  258
    6  236
    7  192
    8  126
    9  71
    10  28
    11  11
    12  7];
time = data(1:12,1);
infected= data(1:12,2);
beta0 = 1;
lamdaa0= 0.019;
lamdas0= 0.0715;
etas0 = 0.03;
etaq0 = 0.04;
gammaa0 = 0.2;
gammaq0 = 0.13;
gammah0 = 0.07;
mua0 = 0.0001;
muh0 = 0.0002;
lb =[0,0,0,0,0,0,0,0,0,0]; ub = [1,1,1,1,1,1,1,1,1,1];
B0 = [beta0; lamdaa0; lamdas0;  etas0; etaq0; gammaa0; gammaq0; gammah0; mua0; muh0 ];
options=optimset('MaxFunEvals', 10000, 'MaxIter', 10000, 'TolFun', 0.0001, 'TolX',0.0001,'Display','on');
[B,resnorm,RESIDUAL,exitflag,OUTPUT,LAMBDA,Jacobian] = lsqcurvefit(@diff1,B0,time,infected,lb,ub,options);
C = diff1(B,time);
plot(time,[infected,C])
function C = diff1(B,time)
x0 =  [1217378052,120,3,2,1,1,0,0];
[t,Cv] = ode45(@DifEq, time, x0);
    function dC = DifEq(t, x)
        N = 1390000000;
        pi = 700 ;
        zetaa = 0.1;
        zetas = 0.2;
        zetaq = 0.3;
        zetah = 0.3;
        omega = 0.2;
        theta = 0.5;
        mu = 0.0000425;
        beta = B(1);
        lamdaa= B(2);
        lamdas = B(3);
        etas = B(4);
        etaq = B(5);
        gammaa = B(6);
        gammaq = B(7);
        gammah = B(8);
        mua = B(9);
        muh = B(10);
        xdot = zeros(7,1);
        
        xdot(1) = pi -beta*(zetaa*x(3) +zetas*x(4) +zetaq*x(5)+zetah*x(6))*(x(1)/N) -mu*x(1);
        xdot(2) = beta*(zetaa*x(3) +zetas*x(4) +zetaq*x(5)+zetah*x(6))*(x(1)/N) -(omega+mu)*x(2); 
        xdot(3) = theta*omega*x(2)-(lamdaa+gammaa+mua+mu)*x(3);
        xdot(4) = (1-theta)*omega*E-(lamdas+etas+mu)*x(4);
        xdot(5) =  lamdaa*Ia+lamdas*Is-(etaq+gammaq+mu)*x(5);
        xdot(6) =   etas*Is+etaq*Q- (gammah+muh+mu)*x(6);
        xdot(7) =  gammaa*x(4) + gammaq*x(5) + gammah*x(6)
        
    end
C = Cv(:,2);
end

[mallela ankamma rao]

sorry for asking again sir

Thank you very much sir .

Answer 2

[Walter Roberson]

infected= data(1:12,2);

You do that assignment once, before any function calls.

You do call

[B,resnorm,RESIDUAL,exitflag,OUTPUT,LAMBDA,Jacobian] = lsqcurvefit(@diff1,B0,time,infected,lb,ub,options);

but there, infected is going to become the ydata for curve fitting purposes; that copy of infected is not going to be passed to any lower level under the name infected

plot(time,infected ,time,Cv)

You are inside function diff1 . infected is not a defined variable there.

Perhaps you were thinking of somehow passing it in to function diff1() so that they could be used for plotting purposes -- that could be done by parameterizing the @diff1 call.

You do some funny things with time vs t in the plot(), but that should be okay... unless, that is, the ode*() call stopped early. It would be better to plot t,Cv than time,Cv in case the full time span does not get covered.

[mallela ankamma rao]

thanks for your reply sir

please run the program and give one line code for graph because i am confused about graph sir

Thank you sir