I hear a lot about the misuse of p-values in research, but my #1 personal pet peeve is the fact that so many commercial statistic softwares include an R² for non-linear regression. Of course, students believe it is valid when the software provides it... h
@Patricio_Quinos @AgricEnvLetters @FernandoMiguez1 @hpoffenb @isuagronomy @UKAgriculture 2- el R2 esta probado que no es un buen criterio de seleccion, particularmente en el caso de modelos no lineales como estos (fijate q los autores indican usar AIC). Sp
@wpforms You are reporting on the R^2 statistic as a measure of goodness-of-fit. However, this statistic is defined for linear models only and is usually inadequate for non-linear models. Reference: https://t.co/F0jySDSEoq
RT @pwk2013: Why we shouldn't be using R^2 for non-linear models #cheminformatics #QSAR #MachineLearning https://t.co/E5bJyU13gC
Beware of only reporting R^2 for your nonlinear models. #QSAR
RT @pwk2013: Why we shouldn't be using R^2 for non-linear models #cheminformatics #QSAR #MachineLearning https://t.co/E5bJyU13gC
Why we shouldn't be using R^2 for non-linear models #cheminformatics #QSAR #MachineLearning https://t.co/E5bJyU13gC
@zoecournia @wpwalters @davidlmobley I have no problem with a correlation statistic being quoted although need to be careful if relationships between quantities are non-linear. This article may be of interest: https://t.co/E5bJyU13gC
@davidlmobley It's about whether the model used to do the prediction is non-linear. I learned about issue in 2015 while trying to persuade JMP to give me R**2 for curves I'd fit. Here's a reference https://t.co/E5bJyU13gC
@krstoffr shouldn't R2 be avoided with nonlinear models? 🤔 (e.g., https://t.co/nGmbEpnIKn)
Be aware: R2 is an inadequate measure for nonlinear models https://t.co/fGJlMtSgs1
Ecology might be explaining less and less http://t.co/qIwU1m3bce But R2 is not the right measure http://t.co/uTfajA6IlX @erikstokstad
An evaluation of R^2 as an inadequate measure for nonlinear models: http://t.co/7Y2AV8c0jT
How bad is r squared as a measure of fit of a nonlinear function? http://t.co/IXUVAw64Il #rstats