Exploratory Modelling

In brolgar: Browse Over Longitudinal Data Graphically and Analytically in R

options(rmarkdown.html_vignette.check_title = FALSE)
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  warning = FALSE,
  message = FALSE
)

library(brolgar)

It can be useful to fit a model to explore your data. One technique is to fit a linear model for each group in a dataset. For example, you could fit a linear model for each key in the data.

brolgar provides a helper function to help with this, called key_slope().

key_slope() returns the intercept and slope estimate for each key, given a linear model formula. We can get the number of observations, and slope information for each individual to identify those that are decreasing over time.

key_slope(wages,ln_wages ~ xp)

We can then join these summaries back to the data:

library(dplyr)
wages_slope <- key_slope(wages,ln_wages ~ xp) %>%
  left_join(wages, by = "id") 

wages_slope

And highlight those individuals with a negative slope using gghighlight:

library(gghighlight)

wages_slope %>% 
  as_tibble() %>% # workaround for gghighlight + tsibble
  ggplot(aes(x = xp, 
             y = ln_wages, 
             group = id)) + 
  geom_line() +
  gghighlight(.slope_xp < 0)

Find keys near other summaries with keys_near()

We might want to further summarise our exploratory modelling by finding those slopes that are near a five number summary values:

summary(wages_slope$.slope_xp)

Finding those groups that are near these values can be surprisingly challenging!

brolgar makes it easier by providing the keys_near() function. You tell it what the key is, what variable you want to summarise by, and then by default it returns those keys near the five number summary. Let's return the keys near the .slope_xp:

wages_slope %>%
  keys_near(key = id,
            var = .slope_xp)

Here it returns the id, the .slope_xp, and the statistic that it was closest to, and what the difference between the slope_xp and the statistic.

You can visualise these summary keys by joining them back to the data:

wages_slope %>%
  keys_near(key = id,
            var = .slope_xp) %>%
  left_join(wages, by = "id") %>%
  ggplot(aes(x = xp,
             y = ln_wages,
             group = id,
             colour = stat)) + 
  geom_line()

You can read more about keys_near() in the Identifying interesting observations vignette.

brolgar documentation built on Feb. 16, 2023, 7:49 p.m.