用R语言写出如下问题的大致代码，考虑世界原油价格周数据，着重考虑价格变化量序列: (a)对于世界原油价格的周变化量序列，建立一个纯ARMA模型. (b)判断在价格变化量序列中是否存在任何季节性。如果有，进行简单的季节调整来消除季节性影响.(c)对于价格改变量的季节调整序列建立一个纯ARMA模型. (d)对于价格改变量的季节调整序列建立一个ARMA-GARCH模型. (e)用最近56个样本点数据进行回测检验来比较前面的ARMA模型和ARMA-GARCH模型。

以下是大致的代码，仅供参考：

# 导入数据
oil_price <- read.csv("oil_price.csv", header = TRUE)

# 计算价格变化量序列
price_diff <- diff(oil_price$price)

# (a) 纯ARMA模型
library(forecast)
arma_model <- Arima(price_diff, order = c(p, d, q), seasonal = FALSE)

# (b) 判断季节性
library(TSA)
seasonal_test <- seas(x = price_diff, period = 52)
if (seasonal_test$p.value < 0.05) {
  # (c) 季节调整
  season_adj <- decompose(x = price_diff, type = "multiplicative")$random
  # (d) 纯ARMA-GARCH模型
  library(rugarch)
  arma_garch_model <- ugarchspec(mean.model = list(armaOrder = c(p, q)), 
                                 variance.model = list(garchOrder = c(1, 1)), 
                                 distribution.model = "sstd")
  arma_garch_fit <- ugarchfit(spec = arma_garch_model, data = season_adj)
} else {
  # (c) 无需季节调整
  season_adj <- price_diff
  # (d) 纯ARMA模型
  arma_garch_fit <- arma_model
}

# (e) 回测检验
n <- length(price_diff)
test_size <- 56
train_data <- price_diff[1:(n - test_size)]
test_data <- price_diff[(n - test_size + 1):n]

# 评估ARMA模型
arma_forecast <- forecast(arma_model, h = test_size)
arma_mse <- mean((arma_forecast$mean - test_data)^2)

# 评估ARMA-GARCH模型
arma_garch_forecast <- ugarchforecast(arma_garch_fit, n.ahead = test_size)
arma_garch_mse <- mean((arma_garch_forecast@forecast$seriesFor[,"mean"] - test_data)^2)

# 输出结果
cat("ARMA模型的MSE:", arma_mse, "\n")
cat("ARMA-GARCH模型的MSE:", arma_garch_mse, "\n")

需要注意的是，这段代码中的 p, d, q 和 garchOrder 都需要根据实际情况进行调整，以得到更好的模型拟合效果。