Ranking the coefficients

Now that we have the coefficients, we can begin to rank each of the categories by increasing trend. Since the results we have obtained so far are contained in embedded lists, which are a bit difficult to work with, we can perform some code manipulation to transform them into a regular data frame, with one row per category, consisting of the category name, coefficient, and coefficient rank:

library(dplyr)
# extract the coefficients part from the model list, and then transpose the
# data frame so that the coefficient appear one per row, rather than 1 per
# column.

xx <- as.data.frame(fitted_models$model)
xx2 <- as.data.frame(t(xx[2, ]))

# The output does not contain the category name, so we will merge it back
# from the original data frame.

xx4 <- cbind(xx2, as.data.frame(fitted_models))[, c(1, 2)]  #only keep the first two columns

# rank the coefficients...