model.md

Thesis - Model

• 1 Load libraries
• 2 First varaible to start with

Load libraries

library(tidyverse)
library(broom)
library(ggfortify)

First varaible to start with

Let’s check which single variable is best to start with.

df_mdl1 <- df %>% 
  select(-c(name, muni_id, budget_approved_capita_log10)) %>% 
  map(~ lm(df$budget_approved_capita_log10 ~ .x, data = df)) %>% 
  map(glance) %>% 
  map("r.squared") %>% 
  as_tibble() %>% 
  pivot_longer(everything() ,names_to = "var", values_to = "r.squared") %>% 
  arrange(desc(r.squared))

Error in UseMethod("select"): no applicable method for 'select' applied to an object of class "function"

df_mdl1

Error in eval(expr, envir, enclos): object 'df_mdl1' not found

Looks like the population variable has the best correlation with the approved SELA budget. Let’s examine how it fairs with linear regression assumptions.

mdl1_pop <- lm(df$budget_approved_capita_log10 ~ df$pop_log10, data = df)

Error in model.frame.default(formula = df$budget_approved_capita_log10 ~ : 'data' must be a data.frame, environment, or list

df %>% 
  ggplot(aes(pop_log10, budget_approved_capita_log10)) +
  geom_point() +
  geom_smooth(method = "lm", se = FALSE)

Error in `ggplot()`:
!   You're passing a function as global data.
  Have you misspelled the `data` argument in `ggplot()`

# autoplot(mdl1_pop)