I have some data that has missing values. I am trying to fit a curve on this data, using Richard’s Model which is \begin{aligned} I(t)=& \frac{K}{\left[1+e^{-r a(t-\tau)}\right] \frac{1}{a}} \end{aligned} where K is fixed and I am trying to estimate a, \tau, and r. So for any time t, I want to b…

add this: idx = findall(!ismissing,ydata) #find the indices of all non missing data in ydata xdata2 = xdata[idx] #xdata but only in the places where ydata is not missing ydata2 = ydata[idx] #all not missing values in ydata

there is LsqFit.jl for just fitting of non linear functions to data. using Pkg #package manager Pkg.add("LsqFit") #adding LsqFit to your Packages using LsqFit #including LsqFit in your enviroment #p = [a,tau,r] #vector, this is the order of parameters im gonna use @. model(t,p,K) = p[1]*K/(1+exp(-p…

Thanks, this is exactly what I needed except it dosn’t work with missing data. # fit @. richards_model(x, p) = 5500 / (1+exp(-p[3]*p[1]*(x-p[2])))^(1/p[1]) p0 = [2, 80, 0.1] xdata = Vector(df.time) ydata = Vector(df.cumcases) fit = curve_fit(richards_model, xdata, ydata, p0) because typeof(ydata)…

Thanks, that did it. There is a function in the source code that checks the health of data. I may submit a PR to check for missing data as well. function check_data_health(xdata, ydata) if any(isinf, xdata) || any(isinf, ydata) || any(isnan, xdata) || any(isnan, ydata) error("Data con…

[image] affans: Thanks, that did it. There is a function in the source code that checks the health of data. I may submit a PR to check for missing data as well. Please do that! I’ll be sure to review it. Ideally we should just support missing values by ignoring the relevant rows, but as a fir…

But let me also tell you the following: I know the LM is very popular for NLS, but in my own experience, BFGS directly on the ssr can be just as good an approach, if not better.

I am looking for similar curve fitting (minimization of SSE) Julia codes. Do you manage to get this working? Thanks.

What’s wrong with the examples here?

This example is perfectly fine. It also showed method to tackle missing data. What I’m missing is the original question which is to translate the R code into Julia equivalent (e.g. using Optim.jl, GLM.jl or JuMP.jl). Pardon me, I am Julia novice. Much of my experiences is from Excel solver. I also …

For those interested, below is an example of SSE minimization using solver in Julia. It can be easily modified for the posted question. using Optim, Gadfly, Cairo # Julia ver. 1.4, Gadfly ver. 1.2.1, Cairo ver. 1.0.2 # y = f(x1, x2) = β_1*x1 + β_2*x2 + β_3 # simply a function # β_1, β_2 & β_3 are…

Estimating parameters of a function for curve fitting

Specific Domains Optimization (Mathematical)

tbeason December 17, 2019, 8:06pm 2

You might try Optim.jl.

Topic		Replies	Views
Parameter estimation in mathematical models of common functions (sine, cos, exp and other special functions) New to Julia package , physics , curve-fitting	15	494	August 28, 2025
Fitting parameters of the model against data New to Julia curve-fitting	3	1850	July 22, 2022
Nonlinear least squares Parameter estimation with one set of data points New to Julia	17	891	July 15, 2022
Alternatives to LsqFit for nonlinear curve fitting? General Usage curve-fitting	12	3338	December 21, 2023
[ANN] GModelFit.jl: a model fitting framework for Julia Package Announcements	1	684	May 24, 2023

Estimating parameters of a function for curve fitting

Related topics