Skip to main content
F
fahmed-31-Visitor
1-Visitor
February 16, 2016
Question

Non linear regression or best fit eqaution

  • February 16, 2016
  • 12 replies
  • 4547 views

Hello everyone,

I hope everyone is doing great. I have three sets of data like:

A= 1,3,5,6,7,8,9

B= 2,3,4,5,6,7,8

C=13,15,16,19,20

These data s are field  raw data and I need to fit it to make a non-linear regression equation. I will be obliged if you can help me out with it

12 replies

F
Fred_Kohlhepp23-Emerald I
23-Emerald I
February 16, 2016

Mathcad will do a multi-variable, polynomial regression.  Look up "regress" in help

F
Fred_Kohlhepp23-Emerald I
23-Emerald I
February 16, 2016

example

R
RichardJ19-Tanzanite
19-Tanzanite
February 17, 2016

There's more than one approach to what you want. One is to do what Fred suggests, and fit a multinomial (which would actually be a linear regression), another would be to use minerr to fit any arbitrary function to the data. To know what the best approach is requires more information than you have provided though. What is the data? Do you have a specific function in mind to fit to the data? Can you post the data and/or function?

F
fahmed-31-VisitorAuthor
1-Visitor
February 17, 2016

Thank you Richard and Fred...... I have attached the data sheet that I am going to work on it. The thing is statistics is a vast subject and to think of an equation is like getting lost in a ocean........I would really appreciate if you can tell me as how to check as which equation fits the data best.....Also I am sending my actual field data that I require to produce using any statistical equation and since i have very little knowledge on stat, can you make a equation out of my data so that in future I can work with any sample of data s. In my data, stress is dependent on moisture and strain so I need to find a non linear regression equation showing the stress is dependent on moisture and strain. I apologize if I am asking too much but can you show me a step by step procedure as which approach should I generally go.

Thanks

F
Fred_Kohlhepp23-Emerald I
23-Emerald I
February 18, 2016

There's one major problem with the data you provided:  there's no variation in moisture.  So the change in stress (or strain) with moisture cannot be examined.

F
fahmed-31-VisitorAuthor
1-Visitor
February 18, 2016

I have attached the graph that I require to make an equation of it.......I need to make a relation between the moisture content, strain and stress......stress being dependent on moisture content and strain.....in the plot you can see that for various moisture contents, I have different stress strain diagrams so I have to quantify a equation that shows the variation of stress with strain and moisture content......I will be taking a statistics course next semester but for now, I have to develop an equation out of it....Thanks

F
Fred_Kohlhepp23-Emerald I
23-Emerald I
February 18, 2016

We can fit the single moisture values you've given as attached.

The example sent before would try to fit a surface thru all of the data.

F
fahmed-31-VisitorAuthor
1-Visitor
February 18, 2016

Thanks Fred......I have done it before using only two parameters but I need to have a function with three parameters that can have a 3D plot of showing stress is dependent on strain and moisture content. The previous file that you gave me has done the curve fitting but I need to find the equation for that curve you gave

Thanks Richard for your reply....you can see the jpg picture attached that has stress strain relations at various moisture content..

F
Fred_Kohlhepp23-Emerald I
23-Emerald I
February 18, 2016

You can unpack the equations for the two=variable fit surface.  Research "polyfit" in help

Terms & ConditionsAccessibility statement