Created
April 26, 2020 16:43
-
-
Save jthomasmock/6289103a53e96884b320e58cd97fab2f to your computer and use it in GitHub Desktop.
Take many variables, fit a model to them all at once, and visualize them all at once
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
library(tidyverse) | |
library(espnscrapeR) | |
library(broom) | |
library(glue) | |
# Get data from espnscrapeR | |
all_off <- 2000:2019 %>% | |
map_dfr(scrape_team_stats_nfl) | |
# Many LMs ---------------------------------------------------------------- | |
# Our goal is to get all of the measures into a `long` form tibble | |
# so that we can fit all of the models at once, and plot it all at once | |
# Offense ----------------------------------------------------------------- | |
# I got the concept for season vs season 2 from Josh Hermsmeyer's 538 article | |
# https://fivethirtyeight.com/features/why-the-nfl-cant-rely-on-defense/ | |
# His code: https://github.com/friscojosh/defensive-metric-stability | |
long_off_stats <- all_off %>% | |
select(team, pts_game:third_pct, season) %>% | |
mutate(season2 = season + 1, | |
pts_game = as.double(pts_game), | |
plays_scrimmage = as.double(plays_scrimmage)) %>% | |
pivot_longer(cols = c(-team, -season, -season2), names_to = "metric", values_to = "value") | |
(off_stability <- long_off_stats %>% | |
inner_join(long_off_stats, by = c("season2" = "season", "team", "metric")) %>% | |
select(everything(), -season2.y) %>% | |
nest(data = c(-metric)) %>% | |
mutate( | |
fit = map(data, ~ lm(value.y ~ value.x, data = .x)), | |
glanced = map(fit, glance) | |
) %>% | |
hoist(glanced, r.squared = "r.squared") %>% | |
arrange(desc(r.squared)) %>% | |
mutate(metric_label = glue::glue("{metric} (R^2 = {round(r.squared, 3)})"))) | |
(off_plot <- long_off_stats %>% | |
inner_join(long_off_stats, by = c("season2" = "season", "team", "metric")) %>% | |
mutate(metric = factor(metric, | |
levels = pull(off_stability, metric), | |
labels = pull(off_stability, metric_label))) %>% | |
ggplot(aes(x = value.x, y = value.y)) + | |
geom_point(color = "black", alpha = 0.5) + | |
geom_smooth(method = "lm", color = "#ff2b4f", se = FALSE) + | |
facet_wrap(~metric, scales = "free") + | |
labs(x = "\nYear Y Metric", | |
y = "Year Y + 1 Metric\n", | |
title = "Offense Stats - 2000-2019", | |
subtitle = "Linear model fit comparing Year and Year + 1", | |
caption = "Plot: @thomas_mock | Data: espnscrapeR") + | |
theme_minimal() + | |
theme(strip.background = element_rect(fill = "black"), | |
strip.text = element_text(face = "bold", color = "white"))) | |
ggsave("plots/off_stability_plot.png", off_plot, height = 10, width = 10, dpi = 450) |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment