# load necessary libraries
library(stargazer)
library(ggplot2)
library(dplyr)

# import dataset
mydata = read.csv('/dataset/Centaline/Centaline_train.csv',header=TRUE) 
head(mydata)

# regression with interaction term
mydata$LogPrice = log(mydata$Transaction_price)
result <- lm(LogPrice ~ Age_of_property * Close_to_MTR, data = mydata)
summary(result)

# regression with fixed effect
result2 <- lm(LogPrice ~ factor(Transaction_month), data = mydata)
summary(result2)

# regression with quadratic form
result3 <- lm(LogPrice ~ Age_of_property + I(Age_of_property^2), data = mydata)
summary(result3)

# table summary 
stargazer(result3, 
          title = "Regression Output", 
          align = TRUE, 
          out = "regression.txt", 
          type = "text")

# visualization
## aggregate the data by Transaction_year and sum the Transaction_price
aggregated_data <- mydata %>%
  group_by(Transaction_year) %>%
  summarise(Transaction_price = sum(Transaction_price))

## create the bar plot with the aggregated data
ggplot(aggregated_data, aes(x = Transaction_year, y = Transaction_price)) +
  geom_bar(stat = "identity") +
  theme_minimal() +
  labs(x = "Transaction Year", y = "Transaction Price")