Further increase data length for seasonal models to 240 points

STL models appear to need more than 10 seasonal cycles. Increased
all seasonal model examples from 120 to 240 data points (20 years
of monthly data) to ensure sufficient cycles for fitting.

Updated in both forecast-models.Rmd and forecastbaselines.Rmd.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

Author: sbfnk

vignettes/forecast-models.Rmd

@@ -301,9 +301,9 @@ historical periods.
 ```{r lsd}
 # Monthly data with seasonality
 set.seed(123)
-months <- 1:120
+months <- 1:240
 seasonal_pattern <- 20 + 10 * sin(2 * pi * months / 12)
-data <- seasonal_pattern + rnorm(120, sd = 2)
+data <- seasonal_pattern + rnorm(240, sd = 2)
 
 # Fit LSD model
 model <- LSDModel(window_width = 3, s = 12) # Average last 3 years
@@ -314,10 +314,10 @@ fc <- forecast(fitted, interval_method = NoInterval(), horizon = 1:12)
 
 # Plot
 plot(data,
-  type = "l", xlim = c(1, 132),
+  type = "l", xlim = c(1, 252),
   main = "LSD Model: Seasonal Forecast"
 )
-lines(121:132, fc$mean, col = "red", lwd = 2)
+lines(241:252, fc$mean, col = "red", lwd = 2)
 ```
 
 **Parameter Guidance:**
@@ -531,10 +531,10 @@ where $T_t$ is trend, $S_t$ is seasonal, $R_t$ is remainder.
 ```{r stl}
 # Generate trend + seasonality
 set.seed(123)
-time <- 1:120 # 10 years of monthly data
+time <- 1:240 # 20 years of monthly data
 trend <- 0.5 * time
 seasonal <- 10 * sin(2 * pi * time / 12)
-data <- 50 + trend + seasonal + rnorm(120, sd = 2)
+data <- 50 + trend + seasonal + rnorm(240, sd = 2)
 
 # Fit STL model
 model <- STLModel(s = 12)
@@ -545,11 +545,11 @@ fc <- forecast(fitted, interval_method = NoInterval(), horizon = 1:12)
 
 # Plot
 plot(data,
-  type = "l", xlim = c(1, 132),
+  type = "l", xlim = c(1, 252),
   main = "STL Model: Trend + Seasonality",
   ylab = "Value"
 )
-lines(121:132, fc$mean, col = "red", lwd = 2)
+lines(241:252, fc$mean, col = "red", lwd = 2)
 ```
 
 **Parameter Guidance:**
@@ -884,10 +884,10 @@ lines(73:84, fc$mean, col = "red", lwd = 2)
 ```{r sales-example}
 # Simulated retail sales: trend + seasonality + promotions
 set.seed(123)
-months <- 1:120
+months <- 1:240
 trend <- 1000 + 20 * months
 seasonal <- 200 * sin(2 * pi * months / 12) # Holiday peaks
-noise <- rnorm(120, sd = 50)
+noise <- rnorm(240, sd = 50)
 data <- trend + seasonal + noise
 
 # Try multiple models

vignettes/forecastbaselines.Rmd

@@ -320,7 +320,7 @@ For data with strong seasonal patterns, use seasonal models:
 ```{r seasonal}
 # Generate seasonal data (monthly, s=12)
 set.seed(789)
-seasonal_data <- 10 + 5 * sin(2 * pi * (1:120) / 12) + rnorm(120)
+seasonal_data <- 10 + 5 * sin(2 * pi * (1:240) / 12) + rnorm(240)
 
 # STL decomposition model
 stl_model <- STLModel(s = 12)