R/sequential.test_fun.R

In dispRity: Measuring Disparity

# #Getting the data function
# set.pair.subsets <- function(subsets_pair, intercept=NULL) {
#     subsets_pair_out <- list()
#     for(element in 1:length(subsets_pair[[1]])) {
#         tmp_list <- lapply(subsets_pair, `[[`, element)
#         #Getting the subsets from the list
#         subsets_pair_out[[element]] <- list.to.table(tmp_list)
#         #Remove subsets column
#         subsets_pair_out[[element]]$subsets <- NULL
#         #Setting the group as binomial
#         subsets_pair_out[[element]]$group <- c(rep(0, length(subsets_pair[[1]][[element]])), rep(1, length(subsets_pair[[2]][[element]])))
#         #Add intercept (if non-null)
#         if(!is.null(intercept)) {
#             #If intercept is a list get the right element!
#             if(class(intercept) == "list") {
#                 subsets_pair_out[[element]]$intercept <- intercept[[element]][[1]]
#             } else {
#                 subsets_pair_out[[element]]$intercept <- intercept
#             }
#         }
#     }
#     return(subsets_pair_out)
# }
# #Estimating intercept function
# intercept.estimate <- function(intercept0, slope) {

#     #Initialising slope variable
#     slope_length <- length(slope)
    
#     if(slope_length > 1) {
#         #First intercept
#         intercept <- intercept0 + slope[1] * 1
#         for(n in 2:slope_length) {
#             intercept <- intercept + slope[n] * 1
#         }
#     } else {
#         intercept <- intercept0 + slope * 1
#     }

#     return(intercept)
# }


# #Sets the intercept0 for a model
# set.intercept0 <- function(first_model) {
#     #If intercept is significant
#     if(summary(first_model)$coefficients[1, 4] < 0.05) {
#         #Set intercept0
#         intercept0 <- coef(first_model)[1]
#     } else {
#         #Else intercept0 is just 0
#         intercept0 <- 0
#     }
#     return(intercept0)
# }

# #Setting the predicted intercept for the next model
# set.intercept.next <- function(one_model, intercept0) {
#     model_summary <- summary(one_model)$coefficients

#     #Check if the model contains an intercept
#     if(dim(model_summary)[1] != 1) {
#         p_value <- model_summary[2, 4]
#         slope <- model_summary[2, 1]
#     } else {
#         p_value <- model_summary[4]
#         slope <- model_summary[1]
#     }

#     if(p_value > 0.05) {
#         #Set slope to 0 if intercept is not significant
#         slope <- 0
#     }

#     #Calculate the next model's intercept
#     intercept_next <- intercept.estimate(intercept0, slope)

#     return(intercept_next)
# }

# #Creating the model function
# create.model <- function(data, family, intercept = NULL, ...) {
#     if(!is.null(intercept)) {
#         #Estimating only the slope in the model
#         if(intercept == "in.data") {
#             #Intercept is present in the data
#             intercept <- unique(data$intercept)
#         } 
#         #Estimate the model using the intercept
#         #model <- glm(group ~ data - 1 + offset(intercept), data = data, family = family, ...) # For binomial
#         model <- glm(data ~ group - 1 + offset(intercept), data = data, family = family, ...) 
#     } else {
#         #Estimating the intercept and the slope in the model
#         #model <- glm(group ~ data, data = data, family = family, ...) # For binomial
#         model <- glm(data ~ group, data = data, family = family, ...)
#     }

#     return(model)
# }