I think it's more convenient to count MCI yourself beforehand and then use geom_col() for your purposes. Reordering is done by fct_reorder
library(tidyverse)
EDAfilter %>%
count(MCI) %>%
ggplot(aes(x = fct_reorder(factor(MCI), n), y = n, fill = factor(MCI))) +
geom_col() +
theme(legend.position="none") +
coord_flip() +
ggtitle("Crime Records by Major Categories")