# Simple R code: data, logistic model, table, and graphs
# Load packages (install if needed)
# install.packages(c("pROC","knitr"))
library(pROC)
library(knitr)
# 1. Example data
set.seed(1)
df <- data.frame(
Age = sample(22:60, 100, replace = TRUE),
Salary = sample(30000:90000, 100, replace = TRUE),
YearsAtJob = sample(1:20, 100, replace = TRUE),
Attrition = sample(c("Yes","No"), 100, replace = TRUE, prob = c(0.3,0.7))
)
df$Attrition <- factor(df$Attrition, levels = c("No","Yes"))
# 2. Train/test split (70/30)
set.seed(123)
train_idx <- sample(seq_len(nrow(df)), size = 0.7 * nrow(df))
train <- df[train_idx, ]
test <- df[-train_idx, ]
# 3. Fit logistic regression
model <- glm(Attrition ~ Age + Salary + YearsAtJob, data = train, family = binomial)
# 4. Predict probabilities and classes on test set
prob <- predict(model, newdata = test, type = "response") # probability of "Yes"
pred_class <- ifelse(prob > 0.5, "Yes", "No")
# 5. Simple results table (first 8 rows)
res <- data.frame(
Age = test$Age,
Salary = test$Salary,
YearsAtJob = test$YearsAtJob,
Actual = test$Attrition,
Predicted = pred_class,
Prob_Yes = round(prob, 3)
)
knitr::kable(head(res, 8), caption = "Test set predictions (first 8 rows)")
Test set predictions (first 8 rows)
| 1 |
25 |
48664 |
11 |
No |
No |
0.272 |
| 2 |
60 |
59244 |
2 |
No |
No |
0.456 |
| 3 |
22 |
87712 |
20 |
Yes |
No |
0.225 |
| 10 |
42 |
69669 |
14 |
Yes |
No |
0.281 |
| 11 |
31 |
49249 |
5 |
Yes |
No |
0.364 |
| 19 |
58 |
50838 |
4 |
No |
No |
0.403 |
| 20 |
55 |
47531 |
16 |
Yes |
No |
0.225 |
| 24 |
41 |
46801 |
20 |
No |
No |
0.174 |
# 6. Plot: predicted probabilities for first 20 test cases
par(mfrow = c(1,2)) # two plots side-by-side
barplot(head(res$Prob_Yes, 20),
names.arg = 1: min(20, nrow(res)),
ylim = c(0,1),
main = "Predicted Probabilities (first 20)",
ylab = "P(Attrition = Yes)")
abline(h = 0.5, col = "red", lty = 2)
# 7. ROC and AUC
roc_obj <- roc(response = test$Attrition, predictor = prob, levels = c("No","Yes"))
plot(roc_obj, main = paste0("ROC curve (AUC = ", round(auc(roc_obj),3), ")"))
par(mfrow = c(1,1))