I am dealing with three continuous predictor variables (corolla tube width, corolla tubedepth, corolla inclination) and a discontinuous response variable (counts of bee visits). I am confused on how to treat this data set and have it analysed because I happen to have linear data (corolla tube width, corolla tubedepth) circular data (corolla inclination). How can I go about this situation??
0
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
Based on what you've described, a Poisson regression (or one of its relatives like Negative Binomial or Zero-Inflated Poisson) is a good candidate here. Poisson regression models count data as a dependent/response variable, meaning that values of the variable have a Poisson distribution. Predictor/explanatory variables can be continuous or a combination of continuous and categorical variables. Poisson is a member of GLM (Generalized Linear Model) family of models, which allow for response variables to have an error distribution other than normal. There are no assumptions in GLM about the distributions of each of the explanatory variables. GLM just assume linear relationship between the transformed response in terms of the link function and the explanatory variables; e.g., for Poisson regression $log(E(Y|X)) = β_{0} + βX$. The log link function here says how the expected value of the response relates to the linear combination/function of explanatory variables.
As for using circular data as a predictor, you can refer to these Q&A's for suggested approaches: Regression using circular variable (hour from 0~23) as predictor, Use of circular predictors in linear regression.
