nlsLoop: Loops through a non-linear model on many different curves to...
In padpadpadpad/TeamPhytoplankton: For studies into thermal adaptation

Description Usage Arguments Details Value Note Author(s) See Also Examples

Fits the best possible model to each of a set of curves using non-linear least squares regression using nlsLM. The best fit is determined using AIC scores.

1 2	nlsLoop(data, model, tries, id_col, param_bds, r2 = "N", supp.errors = "N", AICc = "Y", control, ...)

`data`	Data that includes the rates
`model`	The formula that is usually fed into nlsLM. Make sure there is a y and an x (the column in your dataframe that is the explanatory variable) eg. ln.resp.cor ~ schoolfield.high(ln.c, Ea, Eh, Th, temp=K, Tc = 25)
`tries`	Number of combinations of different starting parameters that are tried on each curve.
`id_col`	The column name that identifies each curve that is to be fitted. Needs to be in speech marks, ' '.
`param_bds`	Upper and lower boundaries for the start parameters. If missing these default to +/- 1e+09. Need to specified as a vector as : c(lower bound param 1, upper bound param 1, lower bound param 2, upper boung param 2 ...)
`r2`	Whether or not you want the quasi rsquared value to be returned. This defaults to no, (so not specifying the argument or r2 = 'N results in no r2 values being returned), to include the r2 values use r2 = 'Y'
`supp.errors`	If supp.errors = 'Y' then no error messages will be shown from the actual nlsLM function, reducing the number of error messages received while the model works through starting parameters from which the model cannot converge. Advised to only be used once it is expected that error messages in the nlsLM function are not important.
`AICc`	Whether or not the small sample AIC should be used. Defaults to 'Y'. Override this using AICc == 'N'. AICc should be used instead of AIC when sample size is small in comparison to the number of estimated parameters (Burnham & Anderson 2002 recommend its use when n / K < 40).
`...`	Extra arguments to pass to nlsLM if necessary.

Returns a dataframe of the best estimated parameter fits for each level of id_col with assOciated AIC score.

Returns a dataframe with the best parameter values for each curve.

Useful additional arguments for nlsLM include: na.action = na.omit,

lower/upper = c() where these represent upper and lower boundaries for parameter estimates

Daniel Padfield

quasi.r2 for details on the calculation of r squared values for non linear models.

nlsLM for details on additional arguments to pass to the nlsLM function.

See AICc in the AICcmodavg package for application of AICc.

data(PI_data)

nlsLoop(PI_data,
         GPP ~ Eilers_PI(Pmax, Iopt, a, I = light),
         tries = 10,
         id_col = 'temp',
         r2 = 'Y',
         supp.errors = 'Y',
         param_bds = c(0,20000, 0,1000, 0, 500))