Hi Alex, I presume you are trying to calculate the ratio between the "length" column and the "width" column? If so, simply type

new.column

I guess what you want to do is apply some simple non-linear regression? In R you could use something like:

nlm1

Thank you both for your answers!

The issue I have is that I have three data columns and to perform a Length-Weight Relation Model I need just two columns, a Weight one and a Length one, but my Length data its made by two columns, so I have this data:

--Carapace Length----------Weight--

--Long--------Wide-----------in mg---

12.3mm-----8.5mm---------86mg

10.5mm-----5.2mm---------79mg

And I need (because the r package I downloaded) this:

Lenght--------------Weight

41.2mm------------146mg

But I don't know if just doing an average of the Length data to have a data set of two columns (Length and Weight) is statistically correct (I'm not sure and I think its not) or its there a "special" formula or in this case a command to perform that.

Hi,

if what you're looking for is to identify the best model relating linear measurements of carapace to weight, then you can:

1. Make a list of all probable models (given litterature and/or logical relationships). For instance :

Weight=b*Long^c

Weight=b*Wide^c

Weight=b(long/Wide)^c

Weight=a+b*Wide

...

2. Create the model sets as a list of models (see http://cran.r-project.org/web/packages/AICcmodavg/AICcmodavg.pdf, p.15)

3. Use the aictab function from AICcmodavg package to choose the best model. This function creates a model selection table based on one of the following information criteria: AIC, AICc, QAIC, QAICc. The table ranks the models based on the selected information criteria and also provides delta AIC and Akaike weights.

furthermore, if you want to use this model as a prdictive tool, then you should first test it for accuracy and bias on an independent dataset.

Cheers,

Denis

Probably calculation of square of carapace as a first step? Then You can correlate two variables: square and weight.

I am not an expert in carapaces, but for me it sounds reasonably...

You can do linear correlations for both the length + weight, and width + weight.

You can then compare that with correlation for the ratio, for example length/width + weight.

In terms of the coding there are a few tricks you could use.

Assuming your data looks something like this.

data

Thank you everyone for the help!

I was checking your answers and reading a bit more and finally I saw that the best thing to do is just perform a Multiple Linear Regression because "statistically" I cannot just average the weight and length to perform a WLR model, So, using MLR I can plot the WEIGHT ~ LONG + WIDE to graph the information, using the code "lm(WEIGHT ~ LONG + WIDE)"

Thank you everybody for your time and advices!