duplisea/bestE: Find the best explanatory variable for a response variable and the best shape constrainted model to describe the relationship

In fisheries, we are often in the search for an explanatory variable for a ecological process but we do not know what the best one is or the most appropriate shape for the relationship. So for example we might be interested in the impact of water temperature on fish population growth rate. We may have 30 different measures of temperature and we are not exactly sure how each of them might impact population growth rate. This package fits relationships between many different explanatory variables and a response variable (univariate, i.e. one at a a time) and then finds the best fitting model to this using shape constrained GAM models. It then selects the best variable and models based on AIC and significance of the smoother term. It is up to the user to determine which explanatory variable and model makes sense with their hypothesis descending the rank of the best fitting variables and models. Multivariate models are not currently implemented but this is a possible extension without too much difficulty, the problem is that if one does not know what is the best variable to describe a process or what is the shape of the functional process (hence using this package) then it is even less likely that they will know how multiple variables interact to describe a process. It is recommended that if you are building a house of cards, best make it a bungalow.

install.packages("remotes")
remotes::install_github("duplisea/bestE")