YanQiFinalProjectPart3
Yan Qi
2025-11-29
Data preprocessing
library(tidyverse)## Warning: package 'tidyverse' was built under R version 4.5.1## Warning: package 'ggplot2' was built under R version 4.5.2## Warning: package 'forcats' was built under R version 4.5.1## ── 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 4.0.0 ✔ tibble 3.2.1
## ✔ lubridate 1.9.4 ✔ tidyr 1.3.1
## ✔ purrr 1.0.4
## ── 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(ggplot2)
library(DataExplorer)## Warning: package 'DataExplorer' was built under R version 4.5.2library(corrplot)## corrplot 0.95 loadedlibrary(GGally)## Warning: package 'GGally' was built under R version 4.5.2library(dplyr)
library(reshape2)##
## Attaching package: 'reshape2'
##
## The following object is masked from 'package:tidyr':
##
## smiths# import dataset
obesity <- read.csv("ObesityDataSet.csv")
head(obesity)## Gender Age Height Weight family_history_with_overweight FAVC FCVC NCP
## 1 Female 21 1.62 64.0 yes no 2 3
## 2 Female 21 1.52 56.0 yes no 3 3
## 3 Male 23 1.80 77.0 yes no 2 3
## 4 Male 27 1.80 87.0 no no 3 3
## 5 Male 22 1.78 89.8 no no 2 1
## 6 Male 29 1.62 53.0 no yes 2 3
## CAEC SMOKE CH2O SCC FAF TUE CALC MTRANS NObeyesdad
## 1 Sometimes no 2 no 0 1 no Public_Transportation Normal
## 2 Sometimes yes 3 yes 3 0 Sometimes Public_Transportation Normal
## 3 Sometimes no 2 no 2 1 Frequently Public_Transportation Normal
## 4 Sometimes no 2 no 2 0 Frequently Walking Overweight
## 5 Sometimes no 2 no 0 0 Sometimes Public_Transportation Overweight
## 6 Sometimes no 2 no 0 0 Sometimes Automobile NormalcolSums(is.na(obesity))## Gender Age
## 0 0
## Height Weight
## 0 0
## family_history_with_overweight FAVC
## 0 0
## FCVC NCP
## 0 0
## CAEC SMOKE
## 0 0
## CH2O SCC
## 0 0
## FAF TUE
## 0 0
## CALC MTRANS
## 0 0
## NObeyesdad
## 0str(obesity)## 'data.frame': 2111 obs. of 17 variables:
## $ Gender : chr "Female" "Female" "Male" "Male" ...
## $ Age : num 21 21 23 27 22 29 23 22 24 22 ...
## $ Height : num 1.62 1.52 1.8 1.8 1.78 1.62 1.5 1.64 1.78 1.72 ...
## $ Weight : num 64 56 77 87 89.8 53 55 53 64 68 ...
## $ family_history_with_overweight: chr "yes" "yes" "yes" "no" ...
## $ FAVC : chr "no" "no" "no" "no" ...
## $ FCVC : num 2 3 2 3 2 2 3 2 3 2 ...
## $ NCP : num 3 3 3 3 1 3 3 3 3 3 ...
## $ CAEC : chr "Sometimes" "Sometimes" "Sometimes" "Sometimes" ...
## $ SMOKE : chr "no" "yes" "no" "no" ...
## $ CH2O : num 2 3 2 2 2 2 2 2 2 2 ...
## $ SCC : chr "no" "yes" "no" "no" ...
## $ FAF : num 0 3 2 2 0 0 1 3 1 1 ...
## $ TUE : num 1 0 1 0 0 0 0 0 1 1 ...
## $ CALC : chr "no" "Sometimes" "Frequently" "Frequently" ...
## $ MTRANS : chr "Public_Transportation" "Public_Transportation" "Public_Transportation" "Walking" ...
## $ NObeyesdad : chr "Normal" "Normal" "Normal" "Overweight" ...# check height, age and weight
boxplot(obesity$Height, main = "Height")
boxplot(obesity$Weight, main = "Weight")
boxplot(obesity$Age, main = "Age")
as there is no extreme data in height and weight, so I can compute BMI directly
library(stringr)
# add new column "BMI" and compute BMI value
obesity <- obesity %>%
mutate(
BMI = round(Weight / (Height^2), 2)
)
head(obesity)## Gender Age Height Weight family_history_with_overweight FAVC FCVC NCP
## 1 Female 21 1.62 64.0 yes no 2 3
## 2 Female 21 1.52 56.0 yes no 3 3
## 3 Male 23 1.80 77.0 yes no 2 3
## 4 Male 27 1.80 87.0 no no 3 3
## 5 Male 22 1.78 89.8 no no 2 1
## 6 Male 29 1.62 53.0 no yes 2 3
## CAEC SMOKE CH2O SCC FAF TUE CALC MTRANS NObeyesdad
## 1 Sometimes no 2 no 0 1 no Public_Transportation Normal
## 2 Sometimes yes 3 yes 3 0 Sometimes Public_Transportation Normal
## 3 Sometimes no 2 no 2 1 Frequently Public_Transportation Normal
## 4 Sometimes no 2 no 2 0 Frequently Walking Overweight
## 5 Sometimes no 2 no 0 0 Sometimes Public_Transportation Overweight
## 6 Sometimes no 2 no 0 0 Sometimes Automobile Normal
## BMI
## 1 24.39
## 2 24.24
## 3 23.77
## 4 26.85
## 5 28.34
## 6 20.20Since BMI assessment criteria differ for individuals under 18 and adults, and my final app is intended solely for adult evaluation, I will remove all data entries for participants under 18.
obesity_adult <- obesity %>%
filter(Age >= 18)
cat("total data", nrow(obesity), "\n")## total data 2111cat("Adult data", nrow(obesity_adult), "\n")## Adult data 1998write.csv(obesity_adult, "ObesityDataSet_adults_only.csv", row.names = FALSE)min(obesity_adult$Age)## [1] 18# classify according to the BMI
obesity_adult <- obesity_adult %>%
mutate(
BMI_Class = case_when(
BMI < 18.5 ~ "Insufficient",
BMI >= 18.5 & BMI < 25 ~ "Normal",
BMI >= 25 & BMI <30 ~ "Overweight",
BMI >= 30~ "Obesity",
TRUE ~ NA_character_
)
)
obesity_adult <- obesity_adult %>%
mutate(
Label_simple = case_when(
str_detect(tolower(NObeyesdad), "insufficient") ~ "Insufficient",
str_detect(tolower(NObeyesdad), "normal") ~ "Normal",
str_detect(tolower(NObeyesdad), "overweight") ~ "Overweight",
str_detect(tolower(NObeyesdad), "obesity|obese") ~ "Obesity",
TRUE ~ NA_character_
)
)
wrong_rows <- obesity_adult %>%
filter(!is.na(BMI_Class) & !is.na(Label_simple) & BMI_Class != Label_simple)
cat("Wrong data：", nrow(wrong_rows), "\n")## Wrong data： 17print(wrong_rows)## Gender Age Height Weight family_history_with_overweight FAVC
## 1 Female 25.00000 1.560000 45.00000 no yes
## 2 Male 20.00000 1.560000 45.00000 no no
## 3 Female 20.22540 1.550648 44.64180 no yes
## 4 Female 29.97045 1.610863 49.51603 yes yes
## 5 Female 22.37800 1.699568 54.98774 yes yes
## 6 Female 32.59313 1.721903 72.74890 yes yes
## 7 Female 23.71259 1.588597 62.33900 no yes
## 8 Female 37.21816 1.593894 63.32063 yes yes
## 9 Female 42.24475 1.768231 75.62931 yes yes
## 10 Female 18.00000 1.498561 55.37651 no yes
## 11 Male 20.31094 1.849425 85.22812 yes yes
## 12 Female 18.54944 1.545196 72.46786 yes no
## 13 Male 46.49186 1.718097 88.60088 yes yes
## 14 Female 18.19832 1.543338 71.79998 yes no
## 15 Female 18.85047 1.550053 72.95180 yes no
## 16 Female 23.00000 1.665199 83.15115 yes yes
## 17 Female 22.89974 1.661715 82.59579 yes yes
## FCVC NCP CAEC SMOKE CH2O SCC FAF TUE CALC
## 1 2.000000 3.000000 Sometimes no 1.000000 no 0.000000 0.000000 Sometimes
## 2 2.000000 3.000000 Sometimes no 2.000000 no 1.000000 1.000000 Sometimes
## 3 3.000000 2.857787 Frequently no 1.000000 no 0.754646 0.000000 Sometimes
## 4 2.059138 3.904858 Frequently no 2.000000 no 0.821977 0.000000 no
## 5 3.000000 3.000000 Frequently no 2.000000 no 0.139808 0.875464 no
## 6 2.000000 3.000000 Sometimes no 1.000000 no 0.000000 1.339232 Sometimes
## 7 2.397280 2.656588 Sometimes no 2.061062 no 1.912981 1.887386 Sometimes
## 8 2.374640 3.000000 Sometimes no 2.000000 no 2.892922 0.480813 Sometimes
## 9 3.000000 2.951837 Sometimes no 2.112032 no 0.378683 0.000000 Sometimes
## 10 2.000000 3.000000 Sometimes no 1.274718 yes 0.129902 0.978574 Sometimes
## 11 2.146598 3.000000 Sometimes no 2.100112 no 1.171160 0.833761 Sometimes
## 12 3.000000 3.014808 Sometimes no 2.000000 no 1.997529 1.000000 Sometimes
## 13 2.129969 3.000000 Sometimes no 1.568035 no 0.870127 0.000000 no
## 14 3.000000 3.087119 Sometimes no 2.000000 no 1.403872 1.000000 no
## 15 3.000000 3.000974 Sometimes no 2.000000 no 2.274248 1.000000 Sometimes
## 16 2.928234 1.458507 Sometimes no 2.777379 no 0.354541 1.707018 no
## 17 1.203754 1.355354 Sometimes no 2.765593 no 0.128342 1.659476 Sometimes
## MTRANS NObeyesdad BMI BMI_Class Label_simple
## 1 Public_Transportation Normal 18.49 Insufficient Normal
## 2 Public_Transportation Normal 18.49 Insufficient Normal
## 3 Public_Transportation Insufficient 18.57 Normal Insufficient
## 4 Public_Transportation Insufficient 19.08 Normal Insufficient
## 5 Public_Transportation Insufficient 19.04 Normal Insufficient
## 6 Automobile Overweight 24.54 Normal Overweight
## 7 Public_Transportation Overweight 24.70 Normal Overweight
## 8 Automobile Overweight 24.92 Normal Overweight
## 9 Automobile Overweight 24.19 Normal Overweight
## 10 Public_Transportation Overweight 24.66 Normal Overweight
## 11 Public_Transportation Overweight 24.92 Normal Overweight
## 12 Public_Transportation Overweight 30.35 Obesity Overweight
## 13 Automobile Overweight 30.02 Obesity Overweight
## 14 Public_Transportation Overweight 30.14 Obesity Overweight
## 15 Public_Transportation Overweight 30.36 Obesity Overweight
## 16 Public_Transportation Obesity 29.99 Overweight Obesity
## 17 Public_Transportation Obesity 29.91 Overweight Obesity# remove the wrong data
obesity_clean <- obesity_adult %>%
filter(!(BMI_Class != Label_simple))
write.csv(obesity_clean, "obesity_clean.csv", row.names = FALSE)feature checking
ggplot(obesity_clean, aes(x = Height, y = Weight, color = Gender)) +
geom_point(alpha = 0.6) +
labs(
x = "Height (meters)",
y = "Weight (kg)",
title = "Height vs Weight by Gender"
) +
theme_minimal()
ggplot(obesity_clean, aes(x = Gender, y = Weight, fill = Gender))+
geom_boxplot() +
ggtitle("Weight Distribution by Gender")+
theme_minimal()
# explore the relationship between features and BMI classification
# numerical correlation heatmap
numeric_vars <- obesity_clean %>%
select_if(is.numeric)
corr_matrix <- cor(numeric_vars)
corrplot(corr_matrix, method = "color", type = "upper", t1.cex = 0.8, number.cex = 0.7)## Warning in text.default(pos.xlabel[, 1], pos.xlabel[, 2], newcolnames, srt =
## tl.srt, : "t1.cex" is not a graphical parameter## Warning in text.default(pos.ylabel[, 1], pos.ylabel[, 2], newrownames, col =
## tl.col, : "t1.cex" is not a graphical parameter## Warning in title(title, ...): "t1.cex" is not a graphical parameter
Based on the heatmap, BMI shows strong correlations only with weight and height, while the other features display relatively weak associations.
# CHI-SQUARE tests for categorical features
categorical_vars <- obesity_clean %>%
select_if(is.character) %>%
select(-NObeyesdad, -Label_simple)
chi_results <- data.frame(
Feature = character(),
p_value = numeric(),
stringsAsFactors = FALSE
)
cat("===== CHI-SQUARE tests（Categorical Features vs BMI_Class） =====\n")## ===== CHI-SQUARE tests（Categorical Features vs BMI_Class） =====for (col in names(categorical_vars)) {
tbl <- table(obesity_clean[[col]], obesity_clean$BMI_Class)
chi <- chisq.test(tbl)
chi_results <- rbind(
chi_results,
data.frame(Feature = col, p_value = chi$p.value)
)
cat("Feature:", col, " p-value =", chi$p.value, "\n")
}## Feature: Gender p-value = 2.978115e-12
## Feature: family_history_with_overweight p-value = 2.714267e-119
## Feature: FAVC p-value = 8.079538e-42
## Feature: CAEC p-value = 6.34903e-156## Warning in chisq.test(tbl): Chi-squared approximation may be incorrect## Feature: SMOKE p-value = 0.003230958
## Feature: SCC p-value = 1.293436e-20## Warning in chisq.test(tbl): Chi-squared approximation may be incorrect## Feature: CALC p-value = 1.036632e-13## Warning in chisq.test(tbl): Chi-squared approximation may be incorrect## Feature: MTRANS p-value = 1.967326e-25
## Feature: BMI_Class p-value = 0cat("\n===== P-Value（p < 0.05） =====\n")##
## ===== P-Value（p < 0.05） =====print(chi_results %>% filter(p_value < 0.05))## Feature p_value
## 1 Gender 2.978115e-12
## 2 family_history_with_overweight 2.714267e-119
## 3 FAVC 8.079538e-42
## 4 CAEC 6.349030e-156
## 5 SMOKE 3.230958e-03
## 6 SCC 1.293436e-20
## 7 CALC 1.036632e-13
## 8 MTRANS 1.967326e-25
## 9 BMI_Class 0.000000e+00# visualization: categorical features vs. BMI
for (col in names(categorical_vars)) {
p <- ggplot(obesity_clean, aes_string(x = col, y = "BMI")) +
geom_violin(fill = "lightblue", alpha = 0.6) +
geom_boxplot(width = 0.1, alpha = 0.7, outlier.color = "red") +
theme_minimal() +
labs(title = paste("BMI by", col), x = col, y = "BMI") +
theme(axis.text.x = element_text(angle = 45, hjust = 1))
}## Warning: `aes_string()` was deprecated in ggplot2 3.0.0.
## ℹ Please use tidy evaluation idioms with `aes()`.
## ℹ See also `vignette("ggplot2-in-packages")` for more information.
## This warning is displayed once every 8 hours.
## Call `lifecycle::last_lifecycle_warnings()` to see where this warning was
## generated.print(p)
chi_summary <- chi_results %>% arrange(p_value)
print(chi_summary)## Feature p_value
## 1 BMI_Class 0.000000e+00
## 2 CAEC 6.349030e-156
## 3 family_history_with_overweight 2.714267e-119
## 4 FAVC 8.079538e-42
## 5 MTRANS 1.967326e-25
## 6 SCC 1.293436e-20
## 7 CALC 1.036632e-13
## 8 Gender 2.978115e-12
## 9 SMOKE 3.230958e-03library(randomForest)## randomForest 4.7-1.2## Type rfNews() to see new features/changes/bug fixes.##
## Attaching package: 'randomForest'## The following object is masked from 'package:dplyr':
##
## combine## The following object is masked from 'package:ggplot2':
##
## marginobesity_clean$BMI_Class <- as.factor(obesity_clean$BMI_Class)
# Random Forest model
rf_model <- randomForest(
BMI_Class ~ .,
data = obesity_clean %>% select(-BMI, -NObeyesdad, -Label_simple),
importance = TRUE,
ntree = 500
)
importance_df <- as.data.frame(importance(rf_model))
importance_df$Feature <- rownames(importance_df)
# MeanDecreaseGini for importance
importance_df <- importance_df %>%
arrange(desc(MeanDecreaseGini))
# visualization
ggplot(importance_df,
aes(x = reorder(Feature, MeanDecreaseGini),
y = MeanDecreaseGini)) +
geom_col(fill = "lightblue") +
coord_flip() +
theme_minimal() +
labs(title = "Feature Importance for BMI Classification",
x = "Feature",
y = "Importance (MeanDecreaseGini)")
# analyze the features’ contribution to BMI classification. As BMI is
calculate by weight and height, so the result is expected.
# XGBoots model
library(xgboost)## Warning: package 'xgboost' was built under R version 4.5.2##
## Attaching package: 'xgboost'## The following object is masked from 'package:dplyr':
##
## slicelibrary(caret)## Loading required package: lattice##
## Attaching package: 'caret'## The following object is masked from 'package:purrr':
##
## liftlibrary(pROC)## Type 'citation("pROC")' for a citation.##
## Attaching package: 'pROC'## The following objects are masked from 'package:stats':
##
## cov, smooth, varlibrary(Matrix)##
## Attaching package: 'Matrix'## The following objects are masked from 'package:tidyr':
##
## expand, pack, unpack# check distribution of each class
set.seed(123)
table(obesity_clean$BMI_Class)##
## Insufficient Normal Obesity Overweight
## 220 260 962 539data distribution is unbalanced.
obesity_df <- obesity_clean %>% mutate_if(is.character, as.factor)
obesity_df <- obesity_df %>% select(-BMI, -NObeyesdad, -Label_simple)
# one-hot encoding
X_full <- model.matrix(~ . - BMI_Class - 1, data = obesity_df)
X_full <- Matrix(X_full, sparse = TRUE)
y_full <- as.numeric(obesity_df$BMI_Class) - 1
classes <- levels(obesity_df$BMI_Class)
num_class <- length(classes)
cat("Classes:", classes, " Num:", num_class, "\n")## Classes: Insufficient Normal Obesity Overweight Num: 4# stratified train/test split
train_idx <- createDataPartition(y_full, p = 0.8, list = FALSE)
test_idx <- setdiff(seq_len(nrow(X_full)), train_idx)
X_train <- X_full[train_idx, ]
X_test <- X_full[test_idx, ]
y_train <- y_full[train_idx]
y_test <- y_full[test_idx]
cat("Train:", length(y_train), " Test:", length(y_test), "\n")## Train: 1586 Test: 395table(y_train) ## y_train
## 0 1 2 3
## 185 206 763 432# compute class-based sample weights
freq <- table(y_train)
total <- length(y_train)
weight_per_class <- as.numeric(total / (num_class * freq))
names(weight_per_class) <- names(freq)
weight_per_class ## 0 1 2 3
## 2.1432432 1.9247573 0.5196592 0.9178241# map to each training instance
train_weights <- weight_per_class[ as.character(y_train) ]
# sanity check
table(y_train, train_weights)## train_weights
## y_train 0.519659239842726 0.917824074074074 1.9247572815534 2.14324324324324
## 0 0 0 0 185
## 1 0 0 206 0
## 2 763 0 0 0
## 3 0 432 0 0# build DMatrix with weights
dtrain <- xgb.DMatrix(data = X_train, label = y_train)
dtest <- xgb.DMatrix(data = X_test, label = y_test)
setinfo(dtrain, "weight", train_weights)## [1] TRUE# xgboost params & training
params <- list(
objective = "multi:softprob",
eval_metric = "mlogloss",
num_class = num_class,
eta = 0.05,
max_depth = 6,
subsample = 0.8,
colsample_bytree = 0.8
)
watchlist <- list(train = dtrain, eval = dtest)
bst <- xgb.train(params = params,
data = dtrain,
nrounds = 1000,
watchlist = watchlist,
early_stopping_rounds = 30,
verbose = 1)## [1] train-mlogloss:1.305884 eval-mlogloss:1.307361
## Multiple eval metrics are present. Will use eval_mlogloss for early stopping.
## Will train until eval_mlogloss hasn't improved in 30 rounds.
##
## [2] train-mlogloss:1.242716 eval-mlogloss:1.257060
## [3] train-mlogloss:1.183840 eval-mlogloss:1.207443
## [4] train-mlogloss:1.123313 eval-mlogloss:1.147494
## [5] train-mlogloss:1.064457 eval-mlogloss:1.091956
## [6] train-mlogloss:1.018452 eval-mlogloss:1.051462
## [7] train-mlogloss:0.977128 eval-mlogloss:1.012791
## [8] train-mlogloss:0.933218 eval-mlogloss:0.969834
## [9] train-mlogloss:0.899515 eval-mlogloss:0.941642
## [10] train-mlogloss:0.857730 eval-mlogloss:0.902630
## [11] train-mlogloss:0.817551 eval-mlogloss:0.863471
## [12] train-mlogloss:0.786637 eval-mlogloss:0.833265
## [13] train-mlogloss:0.749788 eval-mlogloss:0.797725
## [14] train-mlogloss:0.714223 eval-mlogloss:0.763942
## [15] train-mlogloss:0.688215 eval-mlogloss:0.742605
## [16] train-mlogloss:0.662864 eval-mlogloss:0.717678
## [17] train-mlogloss:0.637296 eval-mlogloss:0.693191
## [18] train-mlogloss:0.614818 eval-mlogloss:0.673634
## [19] train-mlogloss:0.587420 eval-mlogloss:0.648291
## [20] train-mlogloss:0.567515 eval-mlogloss:0.630925
## [21] train-mlogloss:0.547602 eval-mlogloss:0.613813
## [22] train-mlogloss:0.525600 eval-mlogloss:0.593427
## [23] train-mlogloss:0.505416 eval-mlogloss:0.573245
## [24] train-mlogloss:0.485806 eval-mlogloss:0.553608
## [25] train-mlogloss:0.466864 eval-mlogloss:0.533977
## [26] train-mlogloss:0.453156 eval-mlogloss:0.522554
## [27] train-mlogloss:0.437298 eval-mlogloss:0.506799
## [28] train-mlogloss:0.422517 eval-mlogloss:0.494378
## [29] train-mlogloss:0.409197 eval-mlogloss:0.482705
## [30] train-mlogloss:0.394362 eval-mlogloss:0.468119
## [31] train-mlogloss:0.379297 eval-mlogloss:0.453627
## [32] train-mlogloss:0.367063 eval-mlogloss:0.442552
## [33] train-mlogloss:0.353809 eval-mlogloss:0.428988
## [34] train-mlogloss:0.340168 eval-mlogloss:0.414947
## [35] train-mlogloss:0.327973 eval-mlogloss:0.402038
## [36] train-mlogloss:0.316667 eval-mlogloss:0.390314
## [37] train-mlogloss:0.305941 eval-mlogloss:0.378862
## [38] train-mlogloss:0.294851 eval-mlogloss:0.367634
## [39] train-mlogloss:0.286120 eval-mlogloss:0.359869
## [40] train-mlogloss:0.276988 eval-mlogloss:0.351278
## [41] train-mlogloss:0.267935 eval-mlogloss:0.342412
## [42] train-mlogloss:0.258912 eval-mlogloss:0.332808
## [43] train-mlogloss:0.251439 eval-mlogloss:0.326074
## [44] train-mlogloss:0.242261 eval-mlogloss:0.316383
## [45] train-mlogloss:0.232949 eval-mlogloss:0.306806
## [46] train-mlogloss:0.225677 eval-mlogloss:0.299353
## [47] train-mlogloss:0.218456 eval-mlogloss:0.291920
## [48] train-mlogloss:0.213328 eval-mlogloss:0.287015
## [49] train-mlogloss:0.207237 eval-mlogloss:0.280611
## [50] train-mlogloss:0.200335 eval-mlogloss:0.273708
## [51] train-mlogloss:0.194656 eval-mlogloss:0.268158
## [52] train-mlogloss:0.189988 eval-mlogloss:0.263385
## [53] train-mlogloss:0.185453 eval-mlogloss:0.259774
## [54] train-mlogloss:0.180563 eval-mlogloss:0.254739
## [55] train-mlogloss:0.175166 eval-mlogloss:0.249007
## [56] train-mlogloss:0.170333 eval-mlogloss:0.244424
## [57] train-mlogloss:0.165247 eval-mlogloss:0.238954
## [58] train-mlogloss:0.160899 eval-mlogloss:0.234460
## [59] train-mlogloss:0.156636 eval-mlogloss:0.229601
## [60] train-mlogloss:0.153542 eval-mlogloss:0.227369
## [61] train-mlogloss:0.149897 eval-mlogloss:0.223288
## [62] train-mlogloss:0.145907 eval-mlogloss:0.219091
## [63] train-mlogloss:0.142468 eval-mlogloss:0.215701
## [64] train-mlogloss:0.139052 eval-mlogloss:0.212204
## [65] train-mlogloss:0.134413 eval-mlogloss:0.206907
## [66] train-mlogloss:0.131635 eval-mlogloss:0.203992
## [67] train-mlogloss:0.128979 eval-mlogloss:0.201449
## [68] train-mlogloss:0.125384 eval-mlogloss:0.197860
## [69] train-mlogloss:0.122329 eval-mlogloss:0.194934
## [70] train-mlogloss:0.118843 eval-mlogloss:0.191538
## [71] train-mlogloss:0.115542 eval-mlogloss:0.188013
## [72] train-mlogloss:0.111901 eval-mlogloss:0.184037
## [73] train-mlogloss:0.108954 eval-mlogloss:0.181100
## [74] train-mlogloss:0.106026 eval-mlogloss:0.177889
## [75] train-mlogloss:0.103642 eval-mlogloss:0.175954
## [76] train-mlogloss:0.101040 eval-mlogloss:0.173122
## [77] train-mlogloss:0.098622 eval-mlogloss:0.170781
## [78] train-mlogloss:0.096735 eval-mlogloss:0.169264
## [79] train-mlogloss:0.094656 eval-mlogloss:0.166697
## [80] train-mlogloss:0.093132 eval-mlogloss:0.165540
## [81] train-mlogloss:0.091156 eval-mlogloss:0.163627
## [82] train-mlogloss:0.089020 eval-mlogloss:0.160943
## [83] train-mlogloss:0.087183 eval-mlogloss:0.159165
## [84] train-mlogloss:0.085415 eval-mlogloss:0.157325
## [85] train-mlogloss:0.083363 eval-mlogloss:0.154936
## [86] train-mlogloss:0.081347 eval-mlogloss:0.152296
## [87] train-mlogloss:0.079221 eval-mlogloss:0.149549
## [88] train-mlogloss:0.076960 eval-mlogloss:0.147337
## [89] train-mlogloss:0.075306 eval-mlogloss:0.145004
## [90] train-mlogloss:0.073715 eval-mlogloss:0.143127
## [91] train-mlogloss:0.072047 eval-mlogloss:0.141073
## [92] train-mlogloss:0.070522 eval-mlogloss:0.139245
## [93] train-mlogloss:0.069268 eval-mlogloss:0.138216
## [94] train-mlogloss:0.067945 eval-mlogloss:0.137008
## [95] train-mlogloss:0.066535 eval-mlogloss:0.135788
## [96] train-mlogloss:0.065120 eval-mlogloss:0.134242
## [97] train-mlogloss:0.063701 eval-mlogloss:0.132267
## [98] train-mlogloss:0.062265 eval-mlogloss:0.130759
## [99] train-mlogloss:0.060717 eval-mlogloss:0.128795
## [100] train-mlogloss:0.059281 eval-mlogloss:0.126924
## [101] train-mlogloss:0.058232 eval-mlogloss:0.125907
## [102] train-mlogloss:0.056823 eval-mlogloss:0.124241
## [103] train-mlogloss:0.055496 eval-mlogloss:0.122680
## [104] train-mlogloss:0.054371 eval-mlogloss:0.121538
## [105] train-mlogloss:0.053224 eval-mlogloss:0.120130
## [106] train-mlogloss:0.052193 eval-mlogloss:0.118939
## [107] train-mlogloss:0.051152 eval-mlogloss:0.117739
## [108] train-mlogloss:0.050063 eval-mlogloss:0.116218
## [109] train-mlogloss:0.049140 eval-mlogloss:0.115105
## [110] train-mlogloss:0.048278 eval-mlogloss:0.113939
## [111] train-mlogloss:0.047209 eval-mlogloss:0.112767
## [112] train-mlogloss:0.046164 eval-mlogloss:0.111852
## [113] train-mlogloss:0.045146 eval-mlogloss:0.110727
## [114] train-mlogloss:0.044389 eval-mlogloss:0.109952
## [115] train-mlogloss:0.043901 eval-mlogloss:0.109632
## [116] train-mlogloss:0.043339 eval-mlogloss:0.109037
## [117] train-mlogloss:0.042406 eval-mlogloss:0.107734
## [118] train-mlogloss:0.041550 eval-mlogloss:0.106599
## [119] train-mlogloss:0.040622 eval-mlogloss:0.105239
## [120] train-mlogloss:0.039917 eval-mlogloss:0.104336
## [121] train-mlogloss:0.039251 eval-mlogloss:0.103739
## [122] train-mlogloss:0.038658 eval-mlogloss:0.103126
## [123] train-mlogloss:0.037891 eval-mlogloss:0.101974
## [124] train-mlogloss:0.037267 eval-mlogloss:0.101172
## [125] train-mlogloss:0.036670 eval-mlogloss:0.100435
## [126] train-mlogloss:0.036333 eval-mlogloss:0.100120
## [127] train-mlogloss:0.035525 eval-mlogloss:0.099507
## [128] train-mlogloss:0.034836 eval-mlogloss:0.098435
## [129] train-mlogloss:0.034479 eval-mlogloss:0.098188
## [130] train-mlogloss:0.033914 eval-mlogloss:0.097800
## [131] train-mlogloss:0.033244 eval-mlogloss:0.096603
## [132] train-mlogloss:0.032752 eval-mlogloss:0.096109
## [133] train-mlogloss:0.032289 eval-mlogloss:0.095504
## [134] train-mlogloss:0.031724 eval-mlogloss:0.094588
## [135] train-mlogloss:0.031215 eval-mlogloss:0.093789
## [136] train-mlogloss:0.030599 eval-mlogloss:0.093037
## [137] train-mlogloss:0.030037 eval-mlogloss:0.092255
## [138] train-mlogloss:0.029554 eval-mlogloss:0.091931
## [139] train-mlogloss:0.029053 eval-mlogloss:0.091242
## [140] train-mlogloss:0.028635 eval-mlogloss:0.090661
## [141] train-mlogloss:0.028254 eval-mlogloss:0.090245
## [142] train-mlogloss:0.027787 eval-mlogloss:0.089734
## [143] train-mlogloss:0.027229 eval-mlogloss:0.088743
## [144] train-mlogloss:0.026703 eval-mlogloss:0.087978
## [145] train-mlogloss:0.026271 eval-mlogloss:0.087201
## [146] train-mlogloss:0.025868 eval-mlogloss:0.086671
## [147] train-mlogloss:0.025480 eval-mlogloss:0.086054
## [148] train-mlogloss:0.025109 eval-mlogloss:0.085626
## [149] train-mlogloss:0.024624 eval-mlogloss:0.084865
## [150] train-mlogloss:0.024199 eval-mlogloss:0.084073
## [151] train-mlogloss:0.023784 eval-mlogloss:0.083425
## [152] train-mlogloss:0.023491 eval-mlogloss:0.082986
## [153] train-mlogloss:0.023217 eval-mlogloss:0.082660
## [154] train-mlogloss:0.022870 eval-mlogloss:0.082205
## [155] train-mlogloss:0.022604 eval-mlogloss:0.082065
## [156] train-mlogloss:0.022282 eval-mlogloss:0.081398
## [157] train-mlogloss:0.021958 eval-mlogloss:0.081151
## [158] train-mlogloss:0.021669 eval-mlogloss:0.080772
## [159] train-mlogloss:0.021314 eval-mlogloss:0.080302
## [160] train-mlogloss:0.021032 eval-mlogloss:0.079852
## [161] train-mlogloss:0.020737 eval-mlogloss:0.079567
## [162] train-mlogloss:0.020457 eval-mlogloss:0.079262
## [163] train-mlogloss:0.020253 eval-mlogloss:0.079065
## [164] train-mlogloss:0.019969 eval-mlogloss:0.078657
## [165] train-mlogloss:0.019622 eval-mlogloss:0.077994
## [166] train-mlogloss:0.019390 eval-mlogloss:0.077799
## [167] train-mlogloss:0.019148 eval-mlogloss:0.077477
## [168] train-mlogloss:0.018802 eval-mlogloss:0.076875
## [169] train-mlogloss:0.018640 eval-mlogloss:0.076649
## [170] train-mlogloss:0.018453 eval-mlogloss:0.076497
## [171] train-mlogloss:0.018144 eval-mlogloss:0.075932
## [172] train-mlogloss:0.017852 eval-mlogloss:0.075347
## [173] train-mlogloss:0.017585 eval-mlogloss:0.074883
## [174] train-mlogloss:0.017401 eval-mlogloss:0.074994
## [175] train-mlogloss:0.017248 eval-mlogloss:0.074909
## [176] train-mlogloss:0.017002 eval-mlogloss:0.074527
## [177] train-mlogloss:0.016817 eval-mlogloss:0.074268
## [178] train-mlogloss:0.016587 eval-mlogloss:0.073887
## [179] train-mlogloss:0.016434 eval-mlogloss:0.073629
## [180] train-mlogloss:0.016343 eval-mlogloss:0.073598
## [181] train-mlogloss:0.016164 eval-mlogloss:0.073476
## [182] train-mlogloss:0.015935 eval-mlogloss:0.072979
## [183] train-mlogloss:0.015788 eval-mlogloss:0.072912
## [184] train-mlogloss:0.015571 eval-mlogloss:0.072660
## [185] train-mlogloss:0.015405 eval-mlogloss:0.072459
## [186] train-mlogloss:0.015215 eval-mlogloss:0.072177
## [187] train-mlogloss:0.015079 eval-mlogloss:0.071919
## [188] train-mlogloss:0.014956 eval-mlogloss:0.071891
## [189] train-mlogloss:0.014821 eval-mlogloss:0.071788
## [190] train-mlogloss:0.014612 eval-mlogloss:0.071509
## [191] train-mlogloss:0.014483 eval-mlogloss:0.071186
## [192] train-mlogloss:0.014362 eval-mlogloss:0.071095
## [193] train-mlogloss:0.014188 eval-mlogloss:0.070921
## [194] train-mlogloss:0.014016 eval-mlogloss:0.070790
## [195] train-mlogloss:0.013866 eval-mlogloss:0.070666
## [196] train-mlogloss:0.013733 eval-mlogloss:0.070440
## [197] train-mlogloss:0.013585 eval-mlogloss:0.070135
## [198] train-mlogloss:0.013472 eval-mlogloss:0.069991
## [199] train-mlogloss:0.013325 eval-mlogloss:0.069508
## [200] train-mlogloss:0.013253 eval-mlogloss:0.069460
## [201] train-mlogloss:0.013105 eval-mlogloss:0.069309
## [202] train-mlogloss:0.012926 eval-mlogloss:0.069179
## [203] train-mlogloss:0.012827 eval-mlogloss:0.069158
## [204] train-mlogloss:0.012729 eval-mlogloss:0.069009
## [205] train-mlogloss:0.012645 eval-mlogloss:0.068952
## [206] train-mlogloss:0.012529 eval-mlogloss:0.068662
## [207] train-mlogloss:0.012435 eval-mlogloss:0.068451
## [208] train-mlogloss:0.012370 eval-mlogloss:0.068453
## [209] train-mlogloss:0.012265 eval-mlogloss:0.068452
## [210] train-mlogloss:0.012137 eval-mlogloss:0.068182
## [211] train-mlogloss:0.012026 eval-mlogloss:0.068000
## [212] train-mlogloss:0.011920 eval-mlogloss:0.067908
## [213] train-mlogloss:0.011824 eval-mlogloss:0.067849
## [214] train-mlogloss:0.011728 eval-mlogloss:0.067868
## [215] train-mlogloss:0.011599 eval-mlogloss:0.067517
## [216] train-mlogloss:0.011502 eval-mlogloss:0.067457
## [217] train-mlogloss:0.011405 eval-mlogloss:0.067331
## [218] train-mlogloss:0.011340 eval-mlogloss:0.067433
## [219] train-mlogloss:0.011280 eval-mlogloss:0.067339
## [220] train-mlogloss:0.011186 eval-mlogloss:0.067162
## [221] train-mlogloss:0.011044 eval-mlogloss:0.066891
## [222] train-mlogloss:0.010996 eval-mlogloss:0.066827
## [223] train-mlogloss:0.010915 eval-mlogloss:0.066666
## [224] train-mlogloss:0.010860 eval-mlogloss:0.066580
## [225] train-mlogloss:0.010770 eval-mlogloss:0.066386
## [226] train-mlogloss:0.010689 eval-mlogloss:0.066274
## [227] train-mlogloss:0.010604 eval-mlogloss:0.066291
## [228] train-mlogloss:0.010497 eval-mlogloss:0.066191
## [229] train-mlogloss:0.010407 eval-mlogloss:0.065835
## [230] train-mlogloss:0.010343 eval-mlogloss:0.065941
## [231] train-mlogloss:0.010277 eval-mlogloss:0.065859
## [232] train-mlogloss:0.010193 eval-mlogloss:0.065583
## [233] train-mlogloss:0.010117 eval-mlogloss:0.065401
## [234] train-mlogloss:0.010068 eval-mlogloss:0.065240
## [235] train-mlogloss:0.009981 eval-mlogloss:0.065001
## [236] train-mlogloss:0.009908 eval-mlogloss:0.064922
## [237] train-mlogloss:0.009859 eval-mlogloss:0.064806
## [238] train-mlogloss:0.009774 eval-mlogloss:0.064629
## [239] train-mlogloss:0.009714 eval-mlogloss:0.064534
## [240] train-mlogloss:0.009649 eval-mlogloss:0.064347
## [241] train-mlogloss:0.009579 eval-mlogloss:0.064343
## [242] train-mlogloss:0.009527 eval-mlogloss:0.064344
## [243] train-mlogloss:0.009473 eval-mlogloss:0.064425
## [244] train-mlogloss:0.009420 eval-mlogloss:0.064386
## [245] train-mlogloss:0.009372 eval-mlogloss:0.064268
## [246] train-mlogloss:0.009337 eval-mlogloss:0.064181
## [247] train-mlogloss:0.009280 eval-mlogloss:0.064238
## [248] train-mlogloss:0.009219 eval-mlogloss:0.064069
## [249] train-mlogloss:0.009138 eval-mlogloss:0.063877
## [250] train-mlogloss:0.009073 eval-mlogloss:0.063901
## [251] train-mlogloss:0.009039 eval-mlogloss:0.063760
## [252] train-mlogloss:0.008978 eval-mlogloss:0.063662
## [253] train-mlogloss:0.008943 eval-mlogloss:0.063649
## [254] train-mlogloss:0.008889 eval-mlogloss:0.063496
## [255] train-mlogloss:0.008840 eval-mlogloss:0.063363
## [256] train-mlogloss:0.008777 eval-mlogloss:0.063178
## [257] train-mlogloss:0.008716 eval-mlogloss:0.063065
## [258] train-mlogloss:0.008674 eval-mlogloss:0.063130
## [259] train-mlogloss:0.008612 eval-mlogloss:0.063027
## [260] train-mlogloss:0.008562 eval-mlogloss:0.062887
## [261] train-mlogloss:0.008525 eval-mlogloss:0.062968
## [262] train-mlogloss:0.008470 eval-mlogloss:0.062904
## [263] train-mlogloss:0.008438 eval-mlogloss:0.062917
## [264] train-mlogloss:0.008395 eval-mlogloss:0.062786
## [265] train-mlogloss:0.008365 eval-mlogloss:0.062775
## [266] train-mlogloss:0.008317 eval-mlogloss:0.062683
## [267] train-mlogloss:0.008276 eval-mlogloss:0.062618
## [268] train-mlogloss:0.008239 eval-mlogloss:0.062710
## [269] train-mlogloss:0.008209 eval-mlogloss:0.062604
## [270] train-mlogloss:0.008177 eval-mlogloss:0.062452
## [271] train-mlogloss:0.008126 eval-mlogloss:0.062329
## [272] train-mlogloss:0.008101 eval-mlogloss:0.062269
## [273] train-mlogloss:0.008062 eval-mlogloss:0.062214
## [274] train-mlogloss:0.008014 eval-mlogloss:0.062054
## [275] train-mlogloss:0.007984 eval-mlogloss:0.061952
## [276] train-mlogloss:0.007949 eval-mlogloss:0.062049
## [277] train-mlogloss:0.007919 eval-mlogloss:0.062060
## [278] train-mlogloss:0.007877 eval-mlogloss:0.061981
## [279] train-mlogloss:0.007851 eval-mlogloss:0.062044
## [280] train-mlogloss:0.007827 eval-mlogloss:0.061959
## [281] train-mlogloss:0.007783 eval-mlogloss:0.061775
## [282] train-mlogloss:0.007748 eval-mlogloss:0.061724
## [283] train-mlogloss:0.007717 eval-mlogloss:0.061699
## [284] train-mlogloss:0.007691 eval-mlogloss:0.061604
## [285] train-mlogloss:0.007657 eval-mlogloss:0.061517
## [286] train-mlogloss:0.007629 eval-mlogloss:0.061509
## [287] train-mlogloss:0.007590 eval-mlogloss:0.061434
## [288] train-mlogloss:0.007558 eval-mlogloss:0.061375
## [289] train-mlogloss:0.007527 eval-mlogloss:0.061313
## [290] train-mlogloss:0.007495 eval-mlogloss:0.061277
## [291] train-mlogloss:0.007463 eval-mlogloss:0.061241
## [292] train-mlogloss:0.007431 eval-mlogloss:0.061309
## [293] train-mlogloss:0.007410 eval-mlogloss:0.061387
## [294] train-mlogloss:0.007377 eval-mlogloss:0.061343
## [295] train-mlogloss:0.007345 eval-mlogloss:0.061331
## [296] train-mlogloss:0.007314 eval-mlogloss:0.061309
## [297] train-mlogloss:0.007274 eval-mlogloss:0.061226
## [298] train-mlogloss:0.007248 eval-mlogloss:0.061125
## [299] train-mlogloss:0.007225 eval-mlogloss:0.061140
## [300] train-mlogloss:0.007191 eval-mlogloss:0.060935
## [301] train-mlogloss:0.007158 eval-mlogloss:0.060915
## [302] train-mlogloss:0.007129 eval-mlogloss:0.060753
## [303] train-mlogloss:0.007101 eval-mlogloss:0.060719
## [304] train-mlogloss:0.007075 eval-mlogloss:0.060734
## [305] train-mlogloss:0.007043 eval-mlogloss:0.060769
## [306] train-mlogloss:0.007015 eval-mlogloss:0.060616
## [307] train-mlogloss:0.006981 eval-mlogloss:0.060458
## [308] train-mlogloss:0.006956 eval-mlogloss:0.060408
## [309] train-mlogloss:0.006936 eval-mlogloss:0.060342
## [310] train-mlogloss:0.006911 eval-mlogloss:0.060349
## [311] train-mlogloss:0.006890 eval-mlogloss:0.060240
## [312] train-mlogloss:0.006874 eval-mlogloss:0.060285
## [313] train-mlogloss:0.006849 eval-mlogloss:0.060281
## [314] train-mlogloss:0.006829 eval-mlogloss:0.060270
## [315] train-mlogloss:0.006798 eval-mlogloss:0.060113
## [316] train-mlogloss:0.006782 eval-mlogloss:0.060073
## [317] train-mlogloss:0.006762 eval-mlogloss:0.060056
## [318] train-mlogloss:0.006734 eval-mlogloss:0.060045
## [319] train-mlogloss:0.006713 eval-mlogloss:0.059939
## [320] train-mlogloss:0.006687 eval-mlogloss:0.059916
## [321] train-mlogloss:0.006665 eval-mlogloss:0.059863
## [322] train-mlogloss:0.006645 eval-mlogloss:0.059861
## [323] train-mlogloss:0.006625 eval-mlogloss:0.059871
## [324] train-mlogloss:0.006604 eval-mlogloss:0.059887
## [325] train-mlogloss:0.006583 eval-mlogloss:0.059897
## [326] train-mlogloss:0.006570 eval-mlogloss:0.059802
## [327] train-mlogloss:0.006547 eval-mlogloss:0.059759
## [328] train-mlogloss:0.006523 eval-mlogloss:0.059741
## [329] train-mlogloss:0.006495 eval-mlogloss:0.059690
## [330] train-mlogloss:0.006472 eval-mlogloss:0.059714
## [331] train-mlogloss:0.006456 eval-mlogloss:0.059646
## [332] train-mlogloss:0.006428 eval-mlogloss:0.059524
## [333] train-mlogloss:0.006412 eval-mlogloss:0.059444
## [334] train-mlogloss:0.006398 eval-mlogloss:0.059432
## [335] train-mlogloss:0.006381 eval-mlogloss:0.059385
## [336] train-mlogloss:0.006354 eval-mlogloss:0.059411
## [337] train-mlogloss:0.006342 eval-mlogloss:0.059446
## [338] train-mlogloss:0.006326 eval-mlogloss:0.059465
## [339] train-mlogloss:0.006303 eval-mlogloss:0.059453
## [340] train-mlogloss:0.006287 eval-mlogloss:0.059503
## [341] train-mlogloss:0.006268 eval-mlogloss:0.059499
## [342] train-mlogloss:0.006242 eval-mlogloss:0.059417
## [343] train-mlogloss:0.006233 eval-mlogloss:0.059429
## [344] train-mlogloss:0.006217 eval-mlogloss:0.059401
## [345] train-mlogloss:0.006203 eval-mlogloss:0.059473
## [346] train-mlogloss:0.006183 eval-mlogloss:0.059406
## [347] train-mlogloss:0.006166 eval-mlogloss:0.059370
## [348] train-mlogloss:0.006149 eval-mlogloss:0.059405
## [349] train-mlogloss:0.006128 eval-mlogloss:0.059416
## [350] train-mlogloss:0.006111 eval-mlogloss:0.059364
## [351] train-mlogloss:0.006085 eval-mlogloss:0.059414
## [352] train-mlogloss:0.006067 eval-mlogloss:0.059453
## [353] train-mlogloss:0.006050 eval-mlogloss:0.059315
## [354] train-mlogloss:0.006037 eval-mlogloss:0.059350
## [355] train-mlogloss:0.006015 eval-mlogloss:0.059368
## [356] train-mlogloss:0.006004 eval-mlogloss:0.059441
## [357] train-mlogloss:0.005992 eval-mlogloss:0.059456
## [358] train-mlogloss:0.005975 eval-mlogloss:0.059376
## [359] train-mlogloss:0.005964 eval-mlogloss:0.059327
## [360] train-mlogloss:0.005952 eval-mlogloss:0.059270
## [361] train-mlogloss:0.005933 eval-mlogloss:0.059163
## [362] train-mlogloss:0.005918 eval-mlogloss:0.059159
## [363] train-mlogloss:0.005905 eval-mlogloss:0.059134
## [364] train-mlogloss:0.005890 eval-mlogloss:0.058997
## [365] train-mlogloss:0.005866 eval-mlogloss:0.058905
## [366] train-mlogloss:0.005849 eval-mlogloss:0.058890
## [367] train-mlogloss:0.005837 eval-mlogloss:0.058827
## [368] train-mlogloss:0.005824 eval-mlogloss:0.058828
## [369] train-mlogloss:0.005806 eval-mlogloss:0.058875
## [370] train-mlogloss:0.005792 eval-mlogloss:0.058941
## [371] train-mlogloss:0.005768 eval-mlogloss:0.058853
## [372] train-mlogloss:0.005753 eval-mlogloss:0.058734
## [373] train-mlogloss:0.005738 eval-mlogloss:0.058761
## [374] train-mlogloss:0.005726 eval-mlogloss:0.058723
## [375] train-mlogloss:0.005712 eval-mlogloss:0.058680
## [376] train-mlogloss:0.005701 eval-mlogloss:0.058730
## [377] train-mlogloss:0.005689 eval-mlogloss:0.058684
## [378] train-mlogloss:0.005681 eval-mlogloss:0.058673
## [379] train-mlogloss:0.005661 eval-mlogloss:0.058580
## [380] train-mlogloss:0.005644 eval-mlogloss:0.058490
## [381] train-mlogloss:0.005628 eval-mlogloss:0.058640
## [382] train-mlogloss:0.005610 eval-mlogloss:0.058617
## [383] train-mlogloss:0.005599 eval-mlogloss:0.058586
## [384] train-mlogloss:0.005588 eval-mlogloss:0.058550
## [385] train-mlogloss:0.005579 eval-mlogloss:0.058513
## [386] train-mlogloss:0.005562 eval-mlogloss:0.058442
## [387] train-mlogloss:0.005549 eval-mlogloss:0.058545
## [388] train-mlogloss:0.005540 eval-mlogloss:0.058444
## [389] train-mlogloss:0.005526 eval-mlogloss:0.058374
## [390] train-mlogloss:0.005509 eval-mlogloss:0.058316
## [391] train-mlogloss:0.005487 eval-mlogloss:0.058211
## [392] train-mlogloss:0.005480 eval-mlogloss:0.058187
## [393] train-mlogloss:0.005468 eval-mlogloss:0.058129
## [394] train-mlogloss:0.005452 eval-mlogloss:0.058043
## [395] train-mlogloss:0.005440 eval-mlogloss:0.057872
## [396] train-mlogloss:0.005432 eval-mlogloss:0.057912
## [397] train-mlogloss:0.005413 eval-mlogloss:0.057844
## [398] train-mlogloss:0.005401 eval-mlogloss:0.057854
## [399] train-mlogloss:0.005390 eval-mlogloss:0.057737
## [400] train-mlogloss:0.005376 eval-mlogloss:0.057672
## [401] train-mlogloss:0.005362 eval-mlogloss:0.057736
## [402] train-mlogloss:0.005353 eval-mlogloss:0.057716
## [403] train-mlogloss:0.005344 eval-mlogloss:0.057736
## [404] train-mlogloss:0.005332 eval-mlogloss:0.057649
## [405] train-mlogloss:0.005312 eval-mlogloss:0.057551
## [406] train-mlogloss:0.005301 eval-mlogloss:0.057554
## [407] train-mlogloss:0.005289 eval-mlogloss:0.057584
## [408] train-mlogloss:0.005281 eval-mlogloss:0.057507
## [409] train-mlogloss:0.005268 eval-mlogloss:0.057446
## [410] train-mlogloss:0.005257 eval-mlogloss:0.057471
## [411] train-mlogloss:0.005249 eval-mlogloss:0.057548
## [412] train-mlogloss:0.005236 eval-mlogloss:0.057595
## [413] train-mlogloss:0.005228 eval-mlogloss:0.057548
## [414] train-mlogloss:0.005219 eval-mlogloss:0.057572
## [415] train-mlogloss:0.005210 eval-mlogloss:0.057498
## [416] train-mlogloss:0.005199 eval-mlogloss:0.057382
## [417] train-mlogloss:0.005190 eval-mlogloss:0.057386
## [418] train-mlogloss:0.005180 eval-mlogloss:0.057341
## [419] train-mlogloss:0.005170 eval-mlogloss:0.057372
## [420] train-mlogloss:0.005161 eval-mlogloss:0.057397
## [421] train-mlogloss:0.005149 eval-mlogloss:0.057323
## [422] train-mlogloss:0.005139 eval-mlogloss:0.057275
## [423] train-mlogloss:0.005129 eval-mlogloss:0.057303
## [424] train-mlogloss:0.005121 eval-mlogloss:0.057284
## [425] train-mlogloss:0.005112 eval-mlogloss:0.057307
## [426] train-mlogloss:0.005100 eval-mlogloss:0.057318
## [427] train-mlogloss:0.005091 eval-mlogloss:0.057277
## [428] train-mlogloss:0.005083 eval-mlogloss:0.057323
## [429] train-mlogloss:0.005074 eval-mlogloss:0.057324
## [430] train-mlogloss:0.005064 eval-mlogloss:0.057362
## [431] train-mlogloss:0.005055 eval-mlogloss:0.057347
## [432] train-mlogloss:0.005044 eval-mlogloss:0.057348
## [433] train-mlogloss:0.005037 eval-mlogloss:0.057338
## [434] train-mlogloss:0.005028 eval-mlogloss:0.057381
## [435] train-mlogloss:0.005019 eval-mlogloss:0.057374
## [436] train-mlogloss:0.005012 eval-mlogloss:0.057422
## [437] train-mlogloss:0.005004 eval-mlogloss:0.057349
## [438] train-mlogloss:0.004999 eval-mlogloss:0.057402
## [439] train-mlogloss:0.004989 eval-mlogloss:0.057341
## [440] train-mlogloss:0.004981 eval-mlogloss:0.057371
## [441] train-mlogloss:0.004974 eval-mlogloss:0.057410
## [442] train-mlogloss:0.004968 eval-mlogloss:0.057419
## [443] train-mlogloss:0.004960 eval-mlogloss:0.057477
## [444] train-mlogloss:0.004952 eval-mlogloss:0.057539
## [445] train-mlogloss:0.004941 eval-mlogloss:0.057460
## [446] train-mlogloss:0.004935 eval-mlogloss:0.057396
## [447] train-mlogloss:0.004924 eval-mlogloss:0.057407
## [448] train-mlogloss:0.004917 eval-mlogloss:0.057422
## [449] train-mlogloss:0.004907 eval-mlogloss:0.057381
## [450] train-mlogloss:0.004898 eval-mlogloss:0.057422
## [451] train-mlogloss:0.004892 eval-mlogloss:0.057488
## [452] train-mlogloss:0.004884 eval-mlogloss:0.057523
## Stopping. Best iteration:
## [422] train-mlogloss:0.005139 eval-mlogloss:0.057275cat("Best iter:", bst$best_iteration, "\n")## Best iter: 422# prediction
pred_probs <- predict(bst, X_test)
pred_mat <- matrix(pred_probs, ncol = num_class, byrow = TRUE)
pred_labels <- max.col(pred_mat) - 1
pred_factor <- factor(classes[pred_labels + 1], levels = classes)
obs_factor <- factor(classes[y_test + 1], levels = classes)# confusion matrix & per-class metrics
conf <- caret::confusionMatrix(pred_factor, obs_factor)
print(conf)## Confusion Matrix and Statistics
##
## Reference
## Prediction Insufficient Normal Obesity Overweight
## Insufficient 34 2 0 0
## Normal 1 51 0 2
## Obesity 0 0 198 0
## Overweight 0 1 1 105
##
## Overall Statistics
##
## Accuracy : 0.9823
## 95% CI : (0.9638, 0.9928)
## No Information Rate : 0.5038
## P-Value [Acc > NIR] : < 2.2e-16
##
## Kappa : 0.9726
##
## Mcnemar's Test P-Value : NA
##
## Statistics by Class:
##
## Class: Insufficient Class: Normal Class: Obesity
## Sensitivity 0.97143 0.9444 0.9950
## Specificity 0.99444 0.9912 1.0000
## Pos Pred Value 0.94444 0.9444 1.0000
## Neg Pred Value 0.99721 0.9912 0.9949
## Prevalence 0.08861 0.1367 0.5038
## Detection Rate 0.08608 0.1291 0.5013
## Detection Prevalence 0.09114 0.1367 0.5013
## Balanced Accuracy 0.98294 0.9678 0.9975
## Class: Overweight
## Sensitivity 0.9813
## Specificity 0.9931
## Pos Pred Value 0.9813
## Neg Pred Value 0.9931
## Prevalence 0.2709
## Detection Rate 0.2658
## Detection Prevalence 0.2709
## Balanced Accuracy 0.9872# compute per-class precision, recall, F1 (and macro F1)
library(MLmetrics)## Warning: package 'MLmetrics' was built under R version 4.5.1##
## Attaching package: 'MLmetrics'## The following objects are masked from 'package:caret':
##
## MAE, RMSE## The following object is masked from 'package:base':
##
## Recallprecision_vec <- recall_vec <- f1_vec <- numeric(num_class)
for(i in seq_len(num_class)){
cls <- classes[i]
y_true_bin <- ifelse(obs_factor == cls, 1, 0)
y_pred_bin <- ifelse(pred_factor == cls, 1, 0)
precision_vec[i] <- ifelse(sum(y_pred_bin)==0, NA, Precision(y_pred_bin, y_true_bin))
recall_vec[i] <- Recall(y_pred_bin, y_true_bin)
f1_vec[i] <- ifelse(is.na(precision_vec[i]) | (precision_vec[i]+recall_vec[i])==0, NA,
2 * precision_vec[i] * recall_vec[i] / (precision_vec[i] + recall_vec[i]))
}
metrics_df <- data.frame(Class = classes,
Precision = round(precision_vec,3),
Recall = round(recall_vec,3),
F1 = round(f1_vec,3))
print(metrics_df)## Class Precision Recall F1
## 1 Insufficient 0.994 0.997 0.996
## 2 Normal 0.991 0.991 0.991
## 3 Obesity 1.000 0.995 0.997
## 4 Overweight 0.993 0.993 0.993macroF1 <- mean(metrics_df$F1, na.rm = TRUE)
cat("Macro F1:", round(macroF1,3), "\n")## Macro F1: 0.994# per-class ROC / AUC (one-vs-rest)
auc_list <- list()
par(mfrow = c(ceiling(num_class/2), 2))
for(i in seq_len(num_class)){
cls <- classes[i]
true_bin <- ifelse(obs_factor == cls, 1, 0)
probs_i <- pred_mat[, i]
if(length(unique(true_bin)) < 2){ next }
roc_obj <- roc(true_bin, probs_i, quiet = TRUE)
auc_val <- auc(roc_obj)
auc_list[[cls]] <- as.numeric(auc_val)
plot(roc_obj, main = paste0("ROC for: ", cls, " (AUC=", round(auc_val,3), ")"))
}
par(mfrow = c(1,1))
print(auc_list)## $Insufficient
## [1] 0.9996825
##
## $Normal
## [1] 0.9985337
##
## $Obesity
## [1] 0.9999744
##
## $Overweight
## [1] 0.998702check the model using 5-fold validation
# test model(5-fold validation)
set.seed(1234)
# Preprocessing
obesity_df <- obesity_clean %>% mutate_if(is.character, as.factor)
# remove columns that leak label or are redundant (adjust if not present)
drop_cols <- intersect(c("BMI","NObeyesdad","Label_simple"), names(obesity_df))
if(length(drop_cols) > 0) obesity_df <- obesity_df %>% select(-all_of(drop_cols))
# one-hot encode
X_full <- model.matrix(~ . - BMI_Class - 1, data = obesity_df)
X_full <- Matrix(X_full, sparse = TRUE)
y_full <- as.numeric(obesity_df$BMI_Class) - 1
classes <- levels(obesity_df$BMI_Class)
num_class <- length(classes)
n <- nrow(X_full)
cat("Classes:", classes, " Num:", num_class, " Samples:", n, "\n")## Classes: Insufficient Normal Obesity Overweight Num: 4 Samples: 1981# create stratified folds
folds <- createFolds(y = y_full, k = 5, list = TRUE, returnTrain = FALSE)
# placeholders to collect results
fold_results <- list()
acc_vec <- numeric(length(folds))
macroF1_vec <- numeric(length(folds))
# CV loop
for(i in seq_along(folds)){
cat("====== Fold", i, "======\n")
test_idx <- folds[[i]]
train_idx <- setdiff(seq_len(n), test_idx)
X_train <- X_full[train_idx, ]
X_test <- X_full[test_idx, ]
y_train <- y_full[train_idx]
y_test <- y_full[test_idx]
cat("Train:", length(y_train), " Test:", length(y_test), "\n")
print(table(y_train))
# compute inverse-frequency class weight on this training fold
freq <- table(y_train)
total <- length(y_train)
weight_per_class <- as.numeric(total / (num_class * freq))
names(weight_per_class) <- names(freq)
cat("Weight per class (fold):\n"); print(weight_per_class)
train_weights <- weight_per_class[ as.character(y_train) ]
# build DMatrix and set weights
dtrain <- xgb.DMatrix(data = X_train, label = y_train)
dtest <- xgb.DMatrix(data = X_test, label = y_test)
setinfo(dtrain, "weight", train_weights)
# params
params <- list(
objective = "multi:softprob",
eval_metric = "mlogloss",
num_class = num_class,
eta = 0.05,
max_depth = 6,
subsample = 0.8,
colsample_bytree = 0.8
)
watchlist <- list(train = dtrain, eval = dtest)
bst <- xgb.train(params = params,
data = dtrain,
nrounds = 1000,
watchlist = watchlist,
early_stopping_rounds = 30,
verbose = 0) # set to 1 if you want training log
cat("Best iter (fold ", i, "):", bst$best_iteration, "\n")
# predict on test fold
pred_probs <- predict(bst, X_test)
pred_mat <- matrix(pred_probs, ncol = num_class, byrow = TRUE)
pred_labels <- max.col(pred_mat) - 1
pred_factor <- factor(classes[pred_labels + 1], levels = classes)
obs_factor <- factor(classes[y_test + 1], levels = classes)
# confusion matrix & accuracy
conf <- caret::confusionMatrix(pred_factor, obs_factor)
print(conf$table)
acc <- as.numeric(conf$overall["Accuracy"])
cat("Accuracy (fold ", i, "):", round(acc,4), "\n")
# per-class Precision/Recall/F1 (one-vs-rest)
precision_vec <- recall_vec <- f1_vec <- numeric(num_class)
for(j in seq_len(num_class)){
cls <- classes[j]
y_true_bin <- ifelse(obs_factor == cls, 1, 0)
y_pred_bin <- ifelse(pred_factor == cls, 1, 0)
if(sum(y_pred_bin) == 0){
precision_vec[j] <- NA
} else {
precision_vec[j] <- Precision(y_pred_bin, y_true_bin)
}
recall_vec[j] <- Recall(y_pred_bin, y_true_bin)
if(is.na(precision_vec[j]) || (precision_vec[j] + recall_vec[j]) == 0){
f1_vec[j] <- NA
} else {
f1_vec[j] <- 2 * precision_vec[j] * recall_vec[j] / (precision_vec[j] + recall_vec[j])
}
}
metrics_df <- data.frame(Class = classes,
Precision = round(precision_vec,3),
Recall = round(recall_vec,3),
F1 = round(f1_vec,3))
print(metrics_df)
macroF1 <- mean(metrics_df$F1, na.rm = TRUE)
cat("Macro F1 (fold ", i, "):", round(macroF1,4), "\n\n")
# save fold results
fold_results[[i]] <- list(conf = conf, metrics = metrics_df, accuracy = acc, macroF1 = macroF1, best_iter = bst$best_iteration)
acc_vec[i] <- acc
macroF1_vec[i] <- macroF1
}## ====== Fold 1 ======
## Train: 1586 Test: 395
## y_train
## 0 1 2 3
## 178 205 771 432
## Weight per class (fold):
## 0 1 2 3
## 2.2275281 1.9341463 0.5142672 0.9178241
## Best iter (fold 1 ): 459
## Reference
## Prediction Insufficient Normal Obesity Overweight
## Insufficient 40 0 0 0
## Normal 2 54 0 0
## Obesity 0 0 191 0
## Overweight 0 1 0 107
## Accuracy (fold 1 ): 0.9924
## Class Precision Recall F1
## 1 Insufficient 1.000 0.994 0.997
## 2 Normal 0.994 0.997 0.996
## 3 Obesity 1.000 1.000 1.000
## 4 Overweight 0.997 1.000 0.998
## Macro F1 (fold 1 ): 0.9978
##
## ====== Fold 2 ======
## Train: 1585 Test: 396
## y_train
## 0 1 2 3
## 173 212 769 431
## Weight per class (fold):
## 0 1 2 3
## 2.2904624 1.8691038 0.5152796 0.9193735
## Best iter (fold 2 ): 888
## Reference
## Prediction Insufficient Normal Obesity Overweight
## Insufficient 46 0 0 0
## Normal 1 47 0 2
## Obesity 0 0 192 0
## Overweight 0 1 1 106
## Accuracy (fold 2 ): 0.9874
## Class Precision Recall F1
## 1 Insufficient 1.000 0.997 0.999
## 2 Normal 0.991 0.997 0.994
## 3 Obesity 1.000 0.995 0.998
## 4 Overweight 0.993 0.993 0.993
## Macro F1 (fold 2 ): 0.996
##
## ====== Fold 3 ======
## Train: 1584 Test: 397
## y_train
## 0 1 2 3
## 174 206 773 431
## Weight per class (fold):
## 0 1 2 3
## 2.2758621 1.9223301 0.5122898 0.9187935
## Best iter (fold 3 ): 439
## Reference
## Prediction Insufficient Normal Obesity Overweight
## Insufficient 44 0 0 0
## Normal 2 52 0 2
## Obesity 0 0 189 1
## Overweight 0 2 0 105
## Accuracy (fold 3 ): 0.9824
## Class Precision Recall F1
## 1 Insufficient 1.000 0.994 0.997
## 2 Normal 0.988 0.994 0.991
## 3 Obesity 0.995 1.000 0.998
## 4 Overweight 0.993 0.990 0.991
## Macro F1 (fold 3 ): 0.9942
##
## ====== Fold 4 ======
## Train: 1584 Test: 397
## y_train
## 0 1 2 3
## 173 207 773 431
## Weight per class (fold):
## 0 1 2 3
## 2.2890173 1.9130435 0.5122898 0.9187935
## Best iter (fold 4 ): 487
## Reference
## Prediction Insufficient Normal Obesity Overweight
## Insufficient 47 0 0 0
## Normal 0 50 0 3
## Obesity 0 0 189 0
## Overweight 0 3 0 105
## Accuracy (fold 4 ): 0.9849
## Class Precision Recall F1
## 1 Insufficient 1.000 1.000 1.000
## 2 Normal 0.991 0.991 0.991
## 3 Obesity 1.000 1.000 1.000
## 4 Overweight 0.990 0.990 0.990
## Macro F1 (fold 4 ): 0.9952
##
## ====== Fold 5 ======
## Train: 1585 Test: 396
## y_train
## 0 1 2 3
## 182 210 762 431
## Weight per class (fold):
## 0 1 2 3
## 2.1771978 1.8869048 0.5200131 0.9193735
## Best iter (fold 5 ): 332
## Reference
## Prediction Insufficient Normal Obesity Overweight
## Insufficient 36 1 0 0
## Normal 2 48 0 5
## Obesity 0 0 197 1
## Overweight 0 1 3 102
## Accuracy (fold 5 ): 0.9672
## Class Precision Recall F1
## 1 Insufficient 0.997 0.994 0.996
## 2 Normal 0.980 0.994 0.987
## 3 Obesity 0.995 0.985 0.990
## 4 Overweight 0.986 0.979 0.983
## Macro F1 (fold 5 ): 0.989# aggregate CV results
cat("CV Summary \n")## CV Summarycat("Accuracy per fold:", round(acc_vec,4), "\n")## Accuracy per fold: 0.9924 0.9874 0.9824 0.9849 0.9672cat("Mean Accuracy:", round(mean(acc_vec),4), " SD:", round(sd(acc_vec),4), "\n")## Mean Accuracy: 0.9828 SD: 0.0095cat("Macro-F1 per fold:", round(macroF1_vec,4), "\n")## Macro-F1 per fold: 0.9978 0.996 0.9942 0.9952 0.989cat("Mean Macro-F1:", round(mean(macroF1_vec, na.rm = TRUE),4), " SD:", round(sd(macroF1_vec, na.rm = TRUE),4), "\n")## Mean Macro-F1: 0.9944 SD: 0.0033# mean per-class F1 across folds
all_f1 <- sapply(fold_results, function(x) x$metrics$F1)
mean_f1_per_class <- rowMeans(all_f1, na.rm = TRUE)
f1_table <- data.frame(Class = classes, Mean_F1 = round(mean_f1_per_class,3))
print(f1_table)## Class Mean_F1
## 1 Insufficient 0.998
## 2 Normal 0.992
## 3 Obesity 0.997
## 4 Overweight 0.991# SHAP
library(shapviz)## Warning: package 'shapviz' was built under R version 4.5.2X_train_mat <- as.matrix(X_train)
sv <- shapviz(
bst,
X = X_train_mat,
X_pred = X_train_mat
)
sv[[1]] ## 'shapviz' object representing 1585 x 24 SHAP matrix. Top lines:
##
## GenderFemale GenderMale Age Height Weight
## [1,] -0.01918555 -0.04026219 -0.005398898 -1.8223714 -1.404736
## [2,] 0.03688017 -0.03063223 0.014316372 0.4167305 -4.108120
## family_history_with_overweightyes FAVCyes FCVC NCP
## [1,] -0.05461422 -0.002471139 0.3563047 -0.17736690
## [2,] -0.02334714 -0.007228931 -0.4679727 -0.09698293
## CAECFrequently CAECno CAECSometimes SMOKEyes CH2O SCCyes
## [1,] 0.3830118 0.02066617 -0.05515741 0.03535179 0.08977139 0.05184976
## [2,] 0.3604885 0.02309405 -0.06014491 0.01061692 -0.07888518 0.04653170
## FAF TUE CALCFrequently CALCno CALCSometimes MTRANSBike
## [1,] -0.38611311 0.09007351 -0.06098965 0.02174658 0.0001872934 0
## [2,] 0.05980391 -0.04828957 -0.06719342 0.02273862 0.0035742861 0
## MTRANSMotorbike MTRANSPublic_Transportation MTRANSWalking
## [1,] 0 -0.0196206253 0.0006068132
## [2,] 0 -0.0008451333 0.0006068132sv[[2]] ## 'shapviz' object representing 1585 x 24 SHAP matrix. Top lines:
##
## GenderFemale GenderMale Age Height Weight
## [1,] -0.04835163 0.01033439 0.11994301 -0.7314767 1.350759
## [2,] -0.25132990 0.05520798 0.02402467 2.6033335 0.142669
## family_history_with_overweightyes FAVCyes FCVC NCP
## [1,] 0.07482897 0.001545759 0.1225174 0.04041113
## [2,] -0.02583305 0.001086622 0.1687990 0.06940047
## CAECFrequently CAECno CAECSometimes SMOKEyes CH2O SCCyes
## [1,] -0.03834416 -0.12269240 0.04401067 -0.04430963 -0.139888 0.00219492
## [2,] -0.01745589 -0.06407286 -0.03193298 -0.06306682 0.220165 -0.01840627
## FAF TUE CALCFrequently CALCno CALCSometimes MTRANSBike
## [1,] 0.5370671 -2.0849569 0.0009316059 -0.01361955 -0.01620919 0
## [2,] 0.3799591 0.4552251 0.0281654857 -0.02004077 -0.02803967 0
## MTRANSMotorbike MTRANSPublic_Transportation MTRANSWalking
## [1,] 0 0.001698886 0.0988561
## [2,] 0 0.010374645 0.1458712sv[[3]] ## 'shapviz' object representing 1585 x 24 SHAP matrix. Top lines:
##
## GenderFemale GenderMale Age Height Weight
## [1,] 0.08115668 -0.01527371 -0.22100155 1.087026 -4.92558
## [2,] 0.12513314 -0.02537715 -0.04475144 -1.713925 -2.43980
## family_history_with_overweightyes FAVCyes FCVC NCP
## [1,] 0.10047571 0.02033791 0.005842878 -0.007859098
## [2,] 0.08154058 0.01294900 -0.029374696 0.013664409
## CAECFrequently CAECno CAECSometimes SMOKEyes CH2O SCCyes
## [1,] -0.06137488 -0.0001300465 0.05622872 0.002189198 0.031434 0
## [2,] -0.07206173 -0.0001129695 0.04931241 0.006883312 -0.103168 0
## FAF TUE CALCFrequently CALCno CALCSometimes MTRANSBike
## [1,] -0.1387543 0.1066503 -0.025705222 -0.01678733 -0.001238827 0
## [2,] -0.1050465 -0.3212476 -0.009374741 0.03435629 0.006248672 0
## MTRANSMotorbike MTRANSPublic_Transportation MTRANSWalking
## [1,] 0 0.002944148 0
## [2,] 0 0.003165144 0sv[[4]] ## 'shapviz' object representing 1585 x 24 SHAP matrix. Top lines:
##
## GenderFemale GenderMale Age Height Weight
## [1,] 0.01525060 -0.006459752 -0.09971264 0.5422618 -3.2787855
## [2,] 0.06456939 -0.009119193 -0.20831451 -0.6421134 0.7408128
## family_history_with_overweightyes FAVCyes FCVC NCP
## [1,] -0.03248628 -0.04487322 -0.42408159 -0.05330720
## [2,] 0.06613126 -0.04125594 0.08723304 -0.06632033
## CAECFrequently CAECno CAECSometimes SMOKEyes CH2O
## [1,] -0.1522136 0.004388724 0.03619101 0.025668416 0.07353806
## [2,] -0.2052898 -0.007618206 0.06121099 0.002909356 -0.01855685
## SCCyes FAF TUE CALCFrequently CALCno CALCSometimes
## [1,] 0.2128339 -0.1253075 0.3094125 0.10577960 -0.03729240 0.0167471
## [2,] 0.1488610 -0.5001416 -0.4415909 0.06264763 -0.05436797 0.0324832
## MTRANSBike MTRANSMotorbike MTRANSPublic_Transportation MTRANSWalking
## [1,] 0 0.013042245 0.0082967458 -0.07044958
## [2,] 0 0.002273976 0.0003405067 -0.16094941print(classes)## [1] "Insufficient" "Normal" "Obesity" "Overweight"idx_overweight <- which(classes == "Overweight")
idx_obesity <- which(classes == "Obesity")
sv_overweight <- sv[[ idx_overweight ]]
sv_obesity <- sv[[ idx_obesity ]]# features importance visualization
# Obesity
sv_importance(sv_obesity, kind = "both")
# overweight
sv_importance(sv_overweight, kind = "both")
# dependence plot
# age
sv_dependence(sv_obesity, "Age")
# family_history_with_overweight
sv_dependence(sv_obesity, "family_history_with_overweightyes")
# CAECFrequently
sv_dependence(sv_obesity, "CAECFrequently")
# TUE
sv_dependence(sv_obesity, "TUE")
sv_force(sv_obesity, row_id = 10)