# Filter the data for the top country to see how their GDP per capita changes over time
gdp_top_country <- gdp_data %>%
  filter(Country == gdp_top_country_NAME)

# Plot the change
gdp_top_country %>%
  autoplot(GDP_cap) +
  labs(title = paste("GDP per Capita Over Time for", gdp_top_country_NAME),
       y = "GDP per Capita",
       x = "Year")

# Filter data for United States GDP
us_gdp <- global_economy %>%
  filter(Country == "United States")

# Plot United States GDP
us_gdp %>%
  autoplot(GDP) +
  labs(title = "United States GDP Over Time", 
       y = "GDP (USD)",
       x = "Year")+
  scale_y_continuous(labels = scales::dollar)

# Log transform United States GDP
us_gdp %>%
  autoplot(log(GDP)) +
  labs(title = "Log-Transformed United States GDP Over Time", 
       y = "Log GDP",
       x = "Year")

# Filter data for Victorian "Bulls, bullocks and steers" slaughter
vic_livestock <- aus_livestock %>%
  filter(State == "Victoria", Animal == "Bulls, bullocks and steers")

# Plot the Victorian slaughter data
vic_livestock %>%
  autoplot(Count) +
  labs(title = "Slaughter of Victorian (Bulls, Bullocks, and Steers) ", 
       y = "Number of Animals Slaughtered",
       x = "Year")+
  scale_y_continuous(labels = scales::comma)

# Plot the Victorian electricity demand
vic_elec %>%
  autoplot(Demand) +
  labs(title = "Victorian Electricity Demand Over Time", 
       y = "Electricity Demand (MW)", 
       x = "Time")

# Plot gas production
aus_production %>%
  autoplot(Gas) +
  labs(title = "Australian Gas Production Over Time", 
       y = "Gas Production", 
       x = "Year")

# Filter out rows where Gas is missing
aus_production_clean <- aus_production %>%
  filter(!is.na(Gas))

# Estimate the best lambda for Box-Cox transformation
lambda <- aus_production_clean %>%
  features(Gas, features = guerrero) %>%
  pull(lambda_guerrero)

# Apply STL decomposition with Box-Cox transformation
aus_production_clean %>%
 autoplot(box_cox(Gas, lambda)) +
  labs(title = "Box-Cox Transformed Australian Gas Production Over Time", 
       y = "Box-Cox Transformed Gas Production", 
       x = "Year")

# Filter for Tobacco data from aus_production
tobacco_data <- aus_production %>%
  select(Tobacco)  %>%
  filter(!is.na(Tobacco))  # Remove missing values

# Plot original data
tobacco_data %>%
  autoplot(Tobacco) +
  labs(title = "Tobacco Production Over Time", 
       y = "Tobacco Production", x = "Year and Quarter")+
  scale_y_continuous(label = scales::comma)

# Estimate lambda using the Guerrero method
lambda_tobacco <- tobacco_data %>%
  features(Tobacco, features = guerrero) %>%
  pull(lambda_guerrero)

# Apply the Box-Cox transformation and plot transformed data
tobacco_data %>%
  mutate(Tobacco_BoxCox = box_cox(Tobacco, lambda_tobacco)) %>%
  autoplot(Tobacco_BoxCox) +
  labs(title = paste("Box-Cox Transformed Tobacco Production (Lambda =", round(lambda_tobacco, 2), ")"),
       y = "Transformed Tobacco Production", x = "Year and Quarter")+
  scale_y_continuous(label = scales::comma)

# Filter for Economy class passengers between Melbourne and Sydney
ansett_data <- ansett %>%
  filter(Class == "Economy", Airports == "MEL-SYD") %>%
  select(Passengers) %>%
  filter(!is.na(Passengers))  # Remove missing values

ansett_data |>
  autoplot(Passengers)+
  labs(title = "Melbourne and Sydney Economy Passengers", y = "Passengers", x = "Year") +
  scale_y_continuous(label = scales::comma)

# Estimate lambda using the Guerrero method
lambda_ansett <- ansett_data %>%
  features(Passengers, features = guerrero) %>%
  pull(lambda_guerrero)

# Apply the Box-Cox transformation and plot transformed data
ansett_data %>%
  mutate(Passengers_BoxCox = box_cox(Passengers, lambda_ansett)) %>%
  autoplot(Passengers_BoxCox) +
  labs(title = paste("Box-Cox Transformed Melbourne and Sydney Economy Passengers (Lambda =", round(lambda_ansett, 2), ")"),
       y = "Transformed Passengers", x = "Year")+
  scale_y_continuous(labels = scales::label_number())

# Filter for pedestrian counts at Southern Cross Station and remove missing values
pedestrian_data <- pedestrian %>%
  filter(Sensor == "Southern Cross Station") %>%
  select(Count) %>%
  filter(!is.na(Count))  # Remove missing values

pedestrian_data |>
  autoplot(Count)+
  labs(title = "Pedestrian counts at Southern Cross Station", y = "Pedestrian", x = "Date") +
  scale_y_continuous(label = scales::comma)

# Estimate lambda using the Guerrero method
lambda_pedestrian <- pedestrian_data %>%
  features(Count, features = guerrero) %>%
  pull(lambda_guerrero)

# Apply the Box-Cox transformation and plot transformed data
pedestrian_data %>%
  mutate(Count_BoxCox = box_cox(Count, lambda_pedestrian)) %>%
  filter(!is.na(Count_BoxCox)) %>%  # Remove any resulting missing values
  autoplot(Count_BoxCox) +
  labs(title = paste("Box-Cox Transformed Pedestrian Counts at Southern Cross Station (Lambda =", round(lambda_pedestrian, 2), ")"),
       y = "Transformed Pedestrian Count", x = "Date")

# Extract the last 5 years (20 quarters) of Gas data
gas <- aus_production %>%
  tail(5 * 4) %>%  # Extract the last 20 observations (5 years of quarterly data)
  select(Gas)

# Plot the time series for the last 5 years of Gas data
gas %>%
  autoplot(Gas) +
  labs(title = "Australian Gas Production For The Last 5 Years", 
       y = "Gas Production (Million Cubic Meters)", 
       x = "Year/Quarter")

gas  %>%
  gg_season(Gas) +
  labs(title = "Seasonal Decomposition of Australian Gas Production For The Last 5 Years", 
       y = "Gas Production (Million Cubic Meters)", 
       x = "Year/Quarter")

# Apply classical decomposition with multiplicative type
decomposition <- gas %>%
  model(classical_decomposition(Gas, type = "multiplicative"))

# Extract and visualize the components
components <- decomposition %>%
  components() %>%
  filter(!is.na(trend))  # Remove rows with missing trend values

# Plot the components of the decomposition
autoplot(components) +
  labs(title = "Multiplicative Decomposition of Gas Production", 
       y = "Gas Production", 
       x = "Year")

# Print out the trend-cycle and seasonal indices
components %>%
  select(trend, season_adjust) %>%
  head()

components(decomposition) %>%
  as_tsibble() %>%
  autoplot(Gas, aes(colour = "Original Gas")) +  # Mapping colour inside aes()
  geom_line(aes(y = season_adjust, colour = "Seasonally Adjusted")) +  # Mapping seasonally adjusted colour
  geom_line(aes(y = trend, colour = "Trend"), linetype = "dashed") +  # Adding trend line with dashed style
  labs(title = "Gas Production: Original, Seasonally Adjusted, and Trend") +
  scale_colour_manual(values = c("Original Gas" = "blue", 
                                 "Seasonally Adjusted" = "green", 
                                 "Trend" = "red")) +  # Custom colors
  guides(colour = guide_legend(title = "Series"))  # Adding a legend

gas$Gas[1] = gas$Gas[1] + 300


gas %>%
  model(classical_decomposition(Gas, type = "multiplicative")) %>%
  components() %>%
  as_tsibble() %>%
  autoplot(Gas, aes(colour = "Original Gas")) + 
  geom_line(aes(y = season_adjust, colour = "Seasonally Adjusted")) + 
  geom_line(aes(y = trend, colour = "Trend"), linetype = "dashed") +  
  labs(title = "Gas Production with an Outlier at the begining: Original, Seasonally Adjusted, and Trend") +
  scale_colour_manual(values = c("Original Gas" = "blue", 
                                 "Seasonally Adjusted" = "green", 
                                 "Trend" = "red")) +  # Custom colors
  guides(colour = guide_legend(title = "Series"))  # Adding a legend

# Select the last 20 observations and introduce an outlier in the middle
gas <- tail(aus_production, 20) %>% select(Gas)

# Find the middle index and introduce the outlier
middle_index <- round(nrow(gas) / 2)
gas$Gas[middle_index] <- gas$Gas[middle_index] + 300  # Add outlier to the middle observation

# Perform classical decomposition and plot with trend
gas %>%
  model(classical_decomposition(Gas, type = "multiplicative")) %>%
  components() %>%
  as_tsibble() %>%
  autoplot(Gas, aes(colour = "Original Gas")) + 
  geom_line(aes(y = season_adjust, colour = "Seasonally Adjusted")) + 
  geom_line(aes(y = trend, colour = "Trend"), linetype = "dashed") +  
  labs(title = "Gas Production with an Outlier in the Middle: Original, Seasonally Adjusted, and Trend") +
  scale_colour_manual(values = c("Original Gas" = "blue", 
                                 "Seasonally Adjusted" = "green", 
                                 "Trend" = "red")) + 
  guides(colour = guide_legend(title = "Series"))  # Adding a legend

# Select the last 20 observations and introduce an outlier at the end
gas <- tail(aus_production, 20) %>% select(Gas)
gas$Gas[20] <- gas$Gas[20] + 300  # Add outlier to the last observation

# Perform classical decomposition and plot with trend
gas %>%
  model(classical_decomposition(Gas, type = "multiplicative")) %>%
  components() %>%
  as_tsibble() %>%
  autoplot(Gas, aes(colour = "Original Gas")) + 
  geom_line(aes(y = season_adjust, colour = "Seasonally Adjusted")) + 
  geom_line(aes(y = trend, colour = "Trend"), linetype = "dashed") +  
  labs(title = "Gas Production with an Outlier at the End: Original, Seasonally Adjusted, and Trend") +
  scale_colour_manual(values = c("Original Gas" = "blue", 
                                 "Seasonally Adjusted" = "green", 
                                 "Trend" = "red")) + 
  guides(colour = guide_legend(title = "Series"))  # Adding a legend