Goal is to to predict CEO departure (ceo_dismissal).
library(tidyverse)
## ── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
## ✔ dplyr 1.1.4 ✔ readr 2.1.5
## ✔ forcats 1.0.0 ✔ stringr 1.5.1
## ✔ ggplot2 3.5.1 ✔ tibble 3.2.1
## ✔ lubridate 1.9.3 ✔ tidyr 1.3.1
## ✔ purrr 1.0.2
## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::lag() masks stats::lag()
## ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errors
library(readr)
library(correlationfunnel)
## ══ correlationfunnel Tip #2 ════════════════════════════════════════════════════
## Clean your NA's prior to using `binarize()`.
## Missing values and cleaning data are critical to getting great correlations. :)
library(tidymodels)
## ── Attaching packages ────────────────────────────────────── tidymodels 1.2.0 ──
## ✔ broom 1.0.6 ✔ rsample 1.2.1
## ✔ dials 1.3.0 ✔ tune 1.2.1
## ✔ infer 1.0.7 ✔ workflows 1.1.4
## ✔ modeldata 1.4.0 ✔ workflowsets 1.1.0
## ✔ parsnip 1.2.1 ✔ yardstick 1.3.2
## ✔ recipes 1.1.1
## ── Conflicts ───────────────────────────────────────── tidymodels_conflicts() ──
## ✖ scales::discard() masks purrr::discard()
## ✖ dplyr::filter() masks stats::filter()
## ✖ recipes::fixed() masks stringr::fixed()
## ✖ dplyr::lag() masks stats::lag()
## ✖ yardstick::spec() masks readr::spec()
## ✖ recipes::step() masks stats::step()
## • Dig deeper into tidy modeling with R at https://www.tmwr.org
library(textrecipes)
library(tidytext)
library(usemodels)
library(xgboost)
##
## Attaching package: 'xgboost'
##
## The following object is masked from 'package:dplyr':
##
## slice
data <- readr::read_csv('https://raw.githubusercontent.com/rfordatascience/tidytuesday/main/data/2021/2021-04-27/departures.csv')
## Rows: 9423 Columns: 19
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr (8): coname, exec_fullname, interim_coceo, still_there, notes, sources...
## dbl (10): dismissal_dataset_id, gvkey, fyear, co_per_rol, departure_code, c...
## dttm (1): leftofc
##
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
data %>% skimr::skim()
| Name | Piped data |
| Number of rows | 9423 |
| Number of columns | 19 |
| _______________________ | |
| Column type frequency: | |
| character | 8 |
| numeric | 10 |
| POSIXct | 1 |
| ________________________ | |
| Group variables | None |
Variable type: character
| skim_variable | n_missing | complete_rate | min | max | empty | n_unique | whitespace |
|---|---|---|---|---|---|---|---|
| coname | 0 | 1.00 | 2 | 30 | 0 | 3860 | 0 |
| exec_fullname | 0 | 1.00 | 5 | 790 | 0 | 8701 | 0 |
| interim_coceo | 9105 | 0.03 | 6 | 7 | 0 | 6 | 0 |
| still_there | 7311 | 0.22 | 3 | 10 | 0 | 77 | 0 |
| notes | 1644 | 0.83 | 5 | 3117 | 0 | 7755 | 0 |
| sources | 1475 | 0.84 | 18 | 1843 | 0 | 7915 | 0 |
| eight_ks | 4499 | 0.52 | 69 | 3884 | 0 | 4914 | 0 |
| _merge | 0 | 1.00 | 11 | 11 | 0 | 1 | 0 |
Variable type: numeric
| skim_variable | n_missing | complete_rate | mean | sd | p0 | p25 | p50 | p75 | p100 | hist |
|---|---|---|---|---|---|---|---|---|---|---|
| dismissal_dataset_id | 0 | 1.00 | 5684.10 | 25005.46 | 1 | 2305.5 | 4593 | 6812.5 | 559044 | ▇▁▁▁▁ |
| gvkey | 0 | 1.00 | 40132.48 | 53921.34 | 1004 | 7337.0 | 14385 | 60900.5 | 328795 | ▇▁▁▁▁ |
| fyear | 0 | 1.00 | 2007.74 | 8.19 | 1987 | 2000.0 | 2008 | 2016.0 | 2020 | ▁▆▅▅▇ |
| co_per_rol | 0 | 1.00 | 25580.22 | 18202.38 | -1 | 8555.5 | 22980 | 39275.5 | 64602 | ▇▆▅▃▃ |
| departure_code | 1667 | 0.82 | 5.20 | 1.53 | 1 | 5.0 | 5 | 7.0 | 9 | ▁▃▇▅▁ |
| ceo_dismissal | 1813 | 0.81 | 0.20 | 0.40 | 0 | 0.0 | 0 | 0.0 | 1 | ▇▁▁▁▂ |
| tenure_no_ceodb | 0 | 1.00 | 1.03 | 0.17 | 0 | 1.0 | 1 | 1.0 | 3 | ▁▇▁▁▁ |
| max_tenure_ceodb | 0 | 1.00 | 1.05 | 0.24 | 1 | 1.0 | 1 | 1.0 | 4 | ▇▁▁▁▁ |
| fyear_gone | 1802 | 0.81 | 2006.64 | 13.63 | 1980 | 2000.0 | 2007 | 2013.0 | 2997 | ▇▁▁▁▁ |
| cik | 245 | 0.97 | 741469.17 | 486551.43 | 1750 | 106413.0 | 857323 | 1050375.8 | 1808065 | ▆▁▇▂▁ |
Variable type: POSIXct
| skim_variable | n_missing | complete_rate | min | max | median | n_unique |
|---|---|---|---|---|---|---|
| leftofc | 1802 | 0.81 | 1981-01-01 | 2998-04-27 | 2006-12-31 | 3627 |
data_clean <- data %>%
# Clean the target
filter(!is.na(ceo_dismissal)) %>%
mutate(ceo_dismissal = if_else(ceo_dismissal == 1, "dismissed", "not_dis")) %>%
mutate(ceo_dismissal = as.factor(ceo_dismissal)) %>%
# Address too many missing values
select(-still_there, -interim_coceo, -eight_ks) %>%
# Remove irrelevant variables
select(-`_merge`, -sources) %>%
# Remove variables that can't be used
select(-departure_code) %>%
# Remove redundant variables
select(-cik, -gvkey, -co_per_rol, -fyear, -leftofc) %>%
# Remove duplicates in the id variable
distinct(dismissal_dataset_id, .keep_all = TRUE) %>%
# Remove 2997 in fyear_gone
filter(fyear_gone < 2025) %>%
# Convert character columns to factors
mutate(across(c(max_tenure_ceodb, tenure_no_ceodb, fyear_gone), as.factor)) %>%
# Convert character columns to factors
mutate(across(where(is.character), as.factor)) %>%
mutate(notes = as.character(notes)) %>%
# Omit missing values
na.omit()
data_clean %>% skimr::skim()
| Name | Piped data |
| Number of rows | 7458 |
| Number of columns | 8 |
| _______________________ | |
| Column type frequency: | |
| character | 1 |
| factor | 6 |
| numeric | 1 |
| ________________________ | |
| Group variables | None |
Variable type: character
| skim_variable | n_missing | complete_rate | min | max | empty | n_unique | whitespace |
|---|---|---|---|---|---|---|---|
| notes | 0 | 1 | 5 | 3117 | 0 | 7448 | 0 |
Variable type: factor
| skim_variable | n_missing | complete_rate | ordered | n_unique | top_counts |
|---|---|---|---|---|---|
| coname | 0 | 1 | FALSE | 3427 | BAR: 8, CLA: 8, FED: 8, NTN: 8 |
| exec_fullname | 0 | 1 | FALSE | 6961 | Joh: 4, Mel: 4, Alb: 3, Ami: 3 |
| ceo_dismissal | 0 | 1 | FALSE | 2 | not: 5976, dis: 1482 |
| tenure_no_ceodb | 0 | 1 | FALSE | 3 | 1: 7274, 2: 177, 3: 7 |
| max_tenure_ceodb | 0 | 1 | FALSE | 4 | 1: 7123, 2: 317, 3: 15, 4: 3 |
| fyear_gone | 0 | 1 | FALSE | 34 | 200: 378, 199: 350, 200: 332, 200: 320 |
Variable type: numeric
| skim_variable | n_missing | complete_rate | mean | sd | p0 | p25 | p50 | p75 | p100 | hist |
|---|---|---|---|---|---|---|---|---|---|---|
| dismissal_dataset_id | 0 | 1 | 5570.24 | 25786.43 | 1 | 2170.25 | 4321.5 | 6575.75 | 559044 | ▇▁▁▁▁ |
data_clean %>% count(ceo_dismissal)
## # A tibble: 2 × 2
## ceo_dismissal n
## <fct> <int>
## 1 dismissed 1482
## 2 not_dis 5976
data_clean %>%
ggplot(aes(ceo_dismissal)) +
geom_bar()
ceo_dismissal vs. fyear_gone
data_clean %>%
ggplot(aes(group = ceo_dismissal, fyear_gone)) +
geom_boxplot()
correlation plot
# Step 1: Binarize
data_binarized <- data_clean %>%
select(-dismissal_dataset_id, -notes) %>%
na.omit() %>%
binarize()
data_binarized %>% glimpse()
## Rows: 7,458
## Columns: 40
## $ coname__BARRICK_GOLD_CORP <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ `coname__-OTHER` <dbl> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, …
## $ exec_fullname__John_W._Rowe <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ `exec_fullname__-OTHER` <dbl> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, …
## $ ceo_dismissal__dismissed <dbl> 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, …
## $ ceo_dismissal__not_dis <dbl> 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, …
## $ tenure_no_ceodb__1 <dbl> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, …
## $ tenure_no_ceodb__2 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ `tenure_no_ceodb__-OTHER` <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ max_tenure_ceodb__1 <dbl> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, …
## $ max_tenure_ceodb__2 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ `max_tenure_ceodb__-OTHER` <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ fyear_gone__1993 <dbl> 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 0, …
## $ fyear_gone__1994 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ fyear_gone__1995 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, …
## $ fyear_gone__1996 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ fyear_gone__1997 <dbl> 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ fyear_gone__1998 <dbl> 0, 1, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, …
## $ fyear_gone__1999 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ fyear_gone__2000 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ fyear_gone__2001 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0, 1, …
## $ fyear_gone__2002 <dbl> 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ fyear_gone__2003 <dbl> 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ fyear_gone__2004 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ fyear_gone__2005 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ fyear_gone__2006 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ fyear_gone__2007 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, …
## $ fyear_gone__2008 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ fyear_gone__2009 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ fyear_gone__2010 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ fyear_gone__2011 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ fyear_gone__2012 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ fyear_gone__2013 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ fyear_gone__2014 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ fyear_gone__2015 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ fyear_gone__2016 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ fyear_gone__2017 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ fyear_gone__2018 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ fyear_gone__2019 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ `fyear_gone__-OTHER` <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
# Step 2: Correlation
data_correlation <- data_binarized %>%
correlate(ceo_dismissal__dismissed)
data_correlation
## # A tibble: 40 × 3
## feature bin correlation
## <fct> <chr> <dbl>
## 1 ceo_dismissal dismissed 1
## 2 ceo_dismissal not_dis -1
## 3 max_tenure_ceodb 1 0.0577
## 4 max_tenure_ceodb 2 -0.0533
## 5 fyear_gone 1999 -0.0390
## 6 fyear_gone 2002 0.0378
## 7 fyear_gone 2003 0.0303
## 8 fyear_gone 2009 0.0292
## 9 fyear_gone 2008 0.0261
## 10 fyear_gone 1997 -0.0255
## # ℹ 30 more rows
# Step 3: Plot
data_correlation %>%
correlationfunnel::plot_correlation_funnel()
## Warning: ggrepel: 28 unlabeled data points (too many overlaps). Consider
## increasing max.overlaps