library(MASS)
library(ggplot2)
## Warning: package 'ggplot2' was built under R version 4.5.3
data <- read.csv("students_adaptability_level_online_education.csv")
head(data)
## Gender Age Education.Level Institution.Type IT.Student Location
## 1 Boy 21-25 University Non Government No Yes
## 2 Girl 21-25 University Non Government No Yes
## 3 Girl 16-20 College Government No Yes
## 4 Girl 11-15 School Non Government No Yes
## 5 Girl 16-20 School Non Government No Yes
## 6 Boy 11-15 School Non Government No Yes
## Load.shedding Financial.Condition Internet.Type Network.Type Class.Duration
## 1 Low Mid Wifi 4G 3-6
## 2 High Mid Mobile Data 4G 1-3
## 3 Low Mid Wifi 4G 1-3
## 4 Low Mid Mobile Data 4G 1-3
## 5 Low Poor Mobile Data 3G 0
## 6 Low Poor Mobile Data 3G 1-3
## Self.Lms Device Adaptivity.Level
## 1 No Tab Moderate
## 2 Yes Mobile Moderate
## 3 No Mobile Moderate
## 4 No Mobile Moderate
## 5 No Mobile Low
## 6 No Mobile Low
str(data)
## 'data.frame': 1205 obs. of 14 variables:
## $ Gender : chr "Boy" "Girl" "Girl" "Girl" ...
## $ Age : chr "21-25" "21-25" "16-20" "11-15" ...
## $ Education.Level : chr "University" "University" "College" "School" ...
## $ Institution.Type : chr "Non Government" "Non Government" "Government" "Non Government" ...
## $ IT.Student : chr "No" "No" "No" "No" ...
## $ Location : chr "Yes" "Yes" "Yes" "Yes" ...
## $ Load.shedding : chr "Low" "High" "Low" "Low" ...
## $ Financial.Condition: chr "Mid" "Mid" "Mid" "Mid" ...
## $ Internet.Type : chr "Wifi" "Mobile Data" "Wifi" "Mobile Data" ...
## $ Network.Type : chr "4G" "4G" "4G" "4G" ...
## $ Class.Duration : chr "3-6" "1-3" "1-3" "1-3" ...
## $ Self.Lms : chr "No" "Yes" "No" "No" ...
## $ Device : chr "Tab" "Mobile" "Mobile" "Mobile" ...
## $ Adaptivity.Level : chr "Moderate" "Moderate" "Moderate" "Moderate" ...
summary(data)
## Gender Age Education.Level Institution.Type
## Length:1205 Length:1205 Length:1205 Length:1205
## Class :character Class :character Class :character Class :character
## Mode :character Mode :character Mode :character Mode :character
## IT.Student Location Load.shedding Financial.Condition
## Length:1205 Length:1205 Length:1205 Length:1205
## Class :character Class :character Class :character Class :character
## Mode :character Mode :character Mode :character Mode :character
## Internet.Type Network.Type Class.Duration Self.Lms
## Length:1205 Length:1205 Length:1205 Length:1205
## Class :character Class :character Class :character Class :character
## Mode :character Mode :character Mode :character Mode :character
## Device Adaptivity.Level
## Length:1205 Length:1205
## Class :character Class :character
## Mode :character Mode :character
data[] <- lapply(data, function(x) {
if(is.character(x)) as.factor(x) else x
})
data$Adaptivity.Level <- factor(data$Adaptivity.Level,
levels = c("Low","Moderate","High"),
ordered = TRUE)
data <- na.omit(data)
set.seed(123)
index <- sample(1:nrow(data), 0.7 * nrow(data))
train_data <- data[index, ]
test_data <- data[-index, ]
model_lda <- lda(Adaptivity.Level ~ ., data = train_data)
print(model_lda)
## Call:
## lda(Adaptivity.Level ~ ., data = train_data)
##
## Prior probabilities of groups:
## Low Moderate High
## 0.40450771 0.51126928 0.08422301
##
## Group means:
## GenderGirl Age11-15 Age16-20 Age21-25 Age26-30 Age6-10
## Low 0.4956012 0.2434018 0.30498534 0.2991202 0.08211144 0.043988270
## Moderate 0.4292343 0.3410673 0.17169374 0.3526682 0.03480278 0.009280742
## High 0.3098592 0.2957746 0.04225352 0.3943662 0.08450704 0.183098592
## Education.LevelSchool Education.LevelUniversity
## Low 0.3607038 0.3870968
## Moderate 0.4663573 0.3967517
## High 0.5070423 0.4788732
## Institution.TypeNon Government IT.StudentYes LocationYes
## Low 0.5131965 0.1994135 0.6510264
## Moderate 0.8213457 0.3132251 0.8491879
## High 0.8309859 0.3098592 0.9295775
## Load.sheddingLow Financial.ConditionPoor Financial.ConditionRich
## Low 0.7888563 0.2668622 0.02052786
## Moderate 0.8654292 0.1392111 0.04176334
## High 0.8732394 0.1971831 0.47887324
## Internet.TypeWifi Network.Type3G Network.Type4G Class.Duration1-3
## Low 0.3988270 0.3958944 0.5747801 0.6158358
## Moderate 0.3944316 0.3457077 0.6473318 0.7378190
## High 0.6760563 0.2112676 0.7887324 0.7887324
## Class.Duration3-6 Self.LmsYes DeviceMobile DeviceTab
## Low 0.08504399 0.1026393 0.9178886 0.002932551
## Moderate 0.24825986 0.2227378 0.8004640 0.039443155
## High 0.21126761 0.2816901 0.6619718 0.028169014
##
## Coefficients of linear discriminants:
## LD1 LD2
## GenderGirl -0.22152884 -0.04872920
## Age11-15 0.38363422 1.08532337
## Age16-20 -0.01084652 1.08220539
## Age21-25 0.69291043 0.86983346
## Age26-30 1.04543505 0.60465440
## Age6-10 1.49120268 2.97395971
## Education.LevelSchool 0.38387891 -0.47879376
## Education.LevelUniversity -0.37879114 0.35094550
## Institution.TypeNon Government 0.68058205 -0.26542216
## IT.StudentYes -0.11378649 -0.38030346
## LocationYes 0.06539147 -0.06017140
## Load.sheddingLow 0.02020918 -0.41095102
## Financial.ConditionPoor -0.21263758 1.08901010
## Financial.ConditionRich 2.65230241 2.72394475
## Internet.TypeWifi -0.35274834 0.67599502
## Network.Type3G 0.47142466 -0.36079547
## Network.Type4G 1.31274536 -0.45498552
## Class.Duration1-3 1.62311508 -1.18259415
## Class.Duration3-6 1.62914098 -2.25688619
## Self.LmsYes 1.09942498 -0.09322434
## DeviceMobile -0.51356065 -0.04599900
## DeviceTab 0.30707452 -0.81258026
##
## Proportion of trace:
## LD1 LD2
## 0.6184 0.3816
pred <- predict(model_lda, test_data)
conf_matrix <- table(Prediksi = pred$class,
Aktual = test_data$Adaptivity.Level)
print(conf_matrix)
## Aktual
## Prediksi Low Moderate High
## Low 95 45 4
## Moderate 42 134 14
## High 2 15 11
accuracy <- mean(as.character(pred$class) == as.character(test_data$Adaptivity.Level))
print(paste("Akurasi:", round(accuracy * 100, 2), "%"))
## [1] "Akurasi: 66.3 %"
plot(model_lda)
ggplot(train_data, aes(x = Age, fill = Adaptivity.Level)) +
geom_bar(position = "fill") +
labs(title = "Proporsi Tingkat Adaptasi Berdasarkan Umur",
x = "Kelompok Umur",
y = "Proporsi",
fill = "Adaptivity Level") +
theme_minimal()
cat("Akurasi Model:", round(accuracy * 100, 2), "%")
## Akurasi Model: 66.3 %