Apply6
Tindra Bergstrand
2024-10-17
Import data
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 errorslibrary(correlationfunnel)## ══ correlationfunnel Tip #3 ════════════════════════════════════════════════════
## Using `binarize()` with data containing many columns or many rows can increase dimensionality substantially.
## Try subsetting your data column-wise or row-wise to avoid creating too many columns.
## You can always make a big problem smaller by sampling. :)library(tidymodels) #for building models## ── Attaching packages ────────────────────────────────────── tidymodels 1.2.0 ──
## ✔ broom 1.0.5 ✔ rsample 1.2.1
## ✔ dials 1.2.1 ✔ 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 1.3.1
## ✔ recipes 1.0.10## Warning: package 'modeldata' was built under R version 4.3.3## ── 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()
## • Use tidymodels_prefer() to resolve common conflicts.library(textrecipes) # For processing string variable
library(tidytext)
library(ggrepel)## Warning: package 'ggrepel' was built under R version 4.3.3members <- readr::read_csv('https://raw.githubusercontent.com/rfordatascience/tidytuesday/master/data/2020/2020-09-22/members.csv')## Rows: 76519 Columns: 21
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr (10): expedition_id, member_id, peak_id, peak_name, season, sex, citizen...
## dbl (5): year, age, highpoint_metres, death_height_metres, injury_height_me...
## lgl (6): hired, success, solo, oxygen_used, died, injured
##
## 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.Clean data
skimr::skim(members)| Name | members |
| Number of rows | 76519 |
| Number of columns | 21 |
| _______________________ | |
| Column type frequency: | |
| character | 10 |
| logical | 6 |
| numeric | 5 |
| ________________________ | |
| Group variables | None |
Variable type: character
| skim_variable | n_missing | complete_rate | min | max | empty | n_unique | whitespace |
|---|---|---|---|---|---|---|---|
| expedition_id | 0 | 1.00 | 9 | 9 | 0 | 10350 | 0 |
| member_id | 0 | 1.00 | 12 | 12 | 0 | 76518 | 0 |
| peak_id | 0 | 1.00 | 4 | 4 | 0 | 391 | 0 |
| peak_name | 15 | 1.00 | 4 | 25 | 0 | 390 | 0 |
| season | 0 | 1.00 | 6 | 7 | 0 | 5 | 0 |
| sex | 2 | 1.00 | 1 | 1 | 0 | 2 | 0 |
| citizenship | 10 | 1.00 | 2 | 23 | 0 | 212 | 0 |
| expedition_role | 21 | 1.00 | 4 | 25 | 0 | 524 | 0 |
| death_cause | 75413 | 0.01 | 3 | 27 | 0 | 12 | 0 |
| injury_type | 74807 | 0.02 | 3 | 27 | 0 | 11 | 0 |
Variable type: logical
| skim_variable | n_missing | complete_rate | mean | count |
|---|---|---|---|---|
| hired | 0 | 1 | 0.21 | FAL: 60788, TRU: 15731 |
| success | 0 | 1 | 0.38 | FAL: 47320, TRU: 29199 |
| solo | 0 | 1 | 0.00 | FAL: 76398, TRU: 121 |
| oxygen_used | 0 | 1 | 0.24 | FAL: 58286, TRU: 18233 |
| died | 0 | 1 | 0.01 | FAL: 75413, TRU: 1106 |
| injured | 0 | 1 | 0.02 | FAL: 74806, TRU: 1713 |
Variable type: numeric
| skim_variable | n_missing | complete_rate | mean | sd | p0 | p25 | p50 | p75 | p100 | hist |
|---|---|---|---|---|---|---|---|---|---|---|
| year | 0 | 1.00 | 2000.36 | 14.78 | 1905 | 1991 | 2004 | 2012 | 2019 | ▁▁▁▃▇ |
| age | 3497 | 0.95 | 37.33 | 10.40 | 7 | 29 | 36 | 44 | 85 | ▁▇▅▁▁ |
| highpoint_metres | 21833 | 0.71 | 7470.68 | 1040.06 | 3800 | 6700 | 7400 | 8400 | 8850 | ▁▁▆▃▇ |
| death_height_metres | 75451 | 0.01 | 6592.85 | 1308.19 | 400 | 5800 | 6600 | 7550 | 8830 | ▁▁▂▇▆ |
| injury_height_metres | 75510 | 0.01 | 7049.91 | 1214.24 | 400 | 6200 | 7100 | 8000 | 8880 | ▁▁▂▇▇ |
Explore data
members1 <- members %>%
# Treat missing values
select(-death_height_metres, -injury_height_metres, -death_cause, -injury_type, -peak_id) %>%
filter(!is.na(age)) %>%
filter(!is.na(highpoint_metres))
factors_vec1 <- members1 %>% select(hired, success, solo, oxygen_used, died, injured) %>% names()
members1_clean <- members1 %>%
# Address factors imported as numeric
mutate(across(all_of(factors_vec1), as.factor))members1 %>% count(died)## # A tibble: 2 × 2
## died n
## <lgl> <int>
## 1 FALSE 51662
## 2 TRUE 744members1 %>%
ggplot(aes(died)) +
geom_bar()
Died vs. age
members1 %>%
ggplot(aes(died, age)) +
geom_boxplot()
Correlation plot
# Step 1: binarze
data_binarized3 <- members1_clean %>%
select(-expedition_id, -member_id) %>%
na.omit() %>%
binarize()
data_binarized3 %>% glimpse()## Rows: 52,383
## Columns: 67
## $ peak_name__Ama_Dablam <dbl> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, …
## $ peak_name__Annapurna_I <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ peak_name__Baruntse <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ peak_name__Cho_Oyu <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ peak_name__Dhaulagiri_I <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ peak_name__Everest <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ peak_name__Himlung_Himal <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ peak_name__Kangchenjunga <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ peak_name__Lhotse <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ peak_name__Makalu <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ peak_name__Manaslu <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ peak_name__Pumori <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ `peak_name__-OTHER` <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ `year__-Inf_1997` <dbl> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, …
## $ year__1997_2007 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ year__2007_2012 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ year__2012_Inf <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ season__Autumn <dbl> 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ season__Spring <dbl> 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, …
## $ season__Winter <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ `season__-OTHER` <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ sex__F <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ sex__M <dbl> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, …
## $ `age__-Inf_29` <dbl> 0, 0, 1, 0, 1, 0, 1, 0, 0, 1, 0, 0, 1, …
## $ age__29_36 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 1, 0, …
## $ age__36_43 <dbl> 1, 1, 0, 1, 0, 1, 0, 1, 0, 0, 0, 0, 0, …
## $ age__43_Inf <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ citizenship__Australia <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ citizenship__Austria <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ citizenship__Canada <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ citizenship__China <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ citizenship__France <dbl> 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ citizenship__Germany <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ citizenship__India <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ citizenship__Italy <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ citizenship__Japan <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ citizenship__Nepal <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ citizenship__New_Zealand <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ citizenship__Poland <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ citizenship__Russia <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ citizenship__S_Korea <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ citizenship__Spain <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ citizenship__Switzerland <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ citizenship__UK <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ citizenship__USA <dbl> 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, …
## $ `citizenship__-OTHER` <dbl> 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, …
## $ expedition_role__Climber <dbl> 0, 0, 1, 1, 1, 1, 1, 0, 0, 1, 0, 1, 0, …
## $ expedition_role__Deputy_Leader <dbl> 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ `expedition_role__H-A_Worker` <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ expedition_role__Leader <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ `expedition_role__-OTHER` <dbl> 0, 1, 0, 0, 0, 0, 0, 1, 1, 0, 1, 0, 1, …
## $ hired__FALSE <dbl> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, …
## $ hired__TRUE <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ `highpoint_metres__-Inf_6750` <dbl> 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ highpoint_metres__6750_7400 <dbl> 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, …
## $ highpoint_metres__7400_8450 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ highpoint_metres__8450_Inf <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ success__FALSE <dbl> 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ success__TRUE <dbl> 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, …
## $ solo__FALSE <dbl> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, …
## $ `solo__-OTHER` <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ oxygen_used__FALSE <dbl> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, …
## $ oxygen_used__TRUE <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ died__FALSE <dbl> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, …
## $ died__TRUE <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
## $ injured__FALSE <dbl> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, …
## $ injured__TRUE <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …# Step 2: correlation
data_correlation2 <- data_binarized3 %>%
correlate(died__TRUE)## Warning: correlate(): [Data Imbalance Detected] Consider sampling to balance the classes more than 5%
## Column with imbalance: died__TRUEdata_correlation2## # A tibble: 67 × 3
## feature bin correlation
## <fct> <chr> <dbl>
## 1 died FALSE -1
## 2 died TRUE 1
## 3 year -Inf_1997 0.0843
## 4 success FALSE 0.0562
## 5 success TRUE -0.0562
## 6 peak_name Annapurna_I 0.0431
## 7 year 2012_Inf -0.0330
## 8 peak_name Ama_Dablam -0.0323
## 9 peak_name Dhaulagiri_I 0.0315
## 10 expedition_role H-A_Worker -0.0309
## # ℹ 57 more rows# Step 3: plot
data_correlation2 %>%
correlationfunnel::plot_correlation_funnel() ## Warning: ggrepel: 32 unlabeled data points (too many overlaps). Consider
## increasing max.overlaps
Model bulidnig
Split data
library(tidymodels)
set.seed(1234)
members1_clean <- members1_clean #%>% sample_n(100)
members_split <- initial_split(members1_clean, strata = died)
members_train <- training(members_split)
members_test <- testing(members_split)
members_cv <- rsample::vfold_cv(members_train, strata = died)
members_cv## # 10-fold cross-validation using stratification
## # A tibble: 10 × 2
## splits id
## <list> <chr>
## 1 <split [35373/3931]> Fold01
## 2 <split [35373/3931]> Fold02
## 3 <split [35373/3931]> Fold03
## 4 <split [35373/3931]> Fold04
## 5 <split [35374/3930]> Fold05
## 6 <split [35374/3930]> Fold06
## 7 <split [35374/3930]> Fold07
## 8 <split [35374/3930]> Fold08
## 9 <split [35374/3930]> Fold09
## 10 <split [35374/3930]> Fold10Preprocess data
library(themis)
xgboost_rec1 <- recipes::recipe(died ~ ., data = members_train) %>%
update_role(member_id, new_role = "ID") %>%
step_other(citizenship, peak_name, expedition_id, threshold = 0.1) %>%
step_dummy(citizenship, peak_name, expedition_id, one_hot = TRUE) %>%
step_impute_knn(all_predictors()) %>%
step_mutate(across(c(member_id, season, sex, expedition_role, hired, success, solo, oxygen_used, injured), as.numeric)) %>%
step_mutate(died = as.factor(died)) %>%
step_smote(died)
xgboost_rec1 %>% prep() %>% juice() %>% glimpse()## Warning: ! There are new levels in a factor: `NA`.## Rows: 77,492
## Columns: 21
## $ member_id <dbl> 4078, 4079, 4080, 4082, 4083, 4081, 4084, 4085…
## $ year <dbl> 1978, 1978, 1978, 1979, 1979, 1979, 1979, 1979…
## $ season <dbl> 1, 1, 1, 2, 2, 2, 2, 2, 2, 2, 2, 1, 1, 1, 1, 1…
## $ sex <dbl> 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2…
## $ age <dbl> 41, 40, 29, 37, 23, 42, 30, 28, 33, 29, 26, 34…
## $ expedition_role <dbl> 107, 114, 27, 27, 27, 89, 129, 27, 27, 76, 129…
## $ hired <dbl> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1…
## $ highpoint_metres <dbl> 6000, 6000, 6000, 6814, 6814, 6814, 6814, 6814…
## $ success <dbl> 1, 1, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2…
## $ solo <dbl> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1…
## $ oxygen_used <dbl> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1…
## $ injured <dbl> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1…
## $ died <fct> FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALS…
## $ citizenship_Nepal <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0…
## $ citizenship_other <dbl> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1…
## $ peak_name_Ama.Dablam <dbl> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1…
## $ peak_name_Cho.Oyu <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0…
## $ peak_name_Everest <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0…
## $ peak_name_other <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0…
## $ expedition_id_HIML13308 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0…
## $ expedition_id_other <dbl> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1…Specify model
library(usemodels)
usemodels::use_xgboost(died ~ ., data = members_train)## xgboost_recipe <-
## recipe(formula = died ~ ., data = members_train) %>%
## step_zv(all_predictors())
##
## xgboost_spec <-
## boost_tree(trees = tune(), min_n = tune(), tree_depth = tune(), learn_rate = tune(),
## loss_reduction = tune(), sample_size = tune()) %>%
## set_mode("classification") %>%
## set_engine("xgboost")
##
## xgboost_workflow <-
## workflow() %>%
## add_recipe(xgboost_recipe) %>%
## add_model(xgboost_spec)
##
## set.seed(56953)
## xgboost_tune <-
## tune_grid(xgboost_workflow, resamples = stop("add your rsample object"), grid = stop("add number of candidate points"))xgboost_spec1 <-
boost_tree(trees = tune(), min_n = tune(), tree_depth = tune(), learn_rate = tune(),
loss_reduction = tune(), sample_size = tune()) %>%
set_mode("classification") %>%
set_engine("xgboost")
xgboost_workflow1 <-
workflow() %>%
add_recipe(xgboost_rec1) %>%
add_model(xgboost_spec1) Tune hyperparameter
doParallel::registerDoParallel()
set.seed(20020)
xgboost_tune <-
tune_grid(xgboost_workflow1,
resamples = members_cv,
grid = 5)## Warning: package 'xgboost' was built under R version 4.3.3