将时间序列分析功能应用于xts对象时出现错误

Question

I want to perform a time series analysis on a daily data stored as xts object. I think not all functions and models that work with ts objects work also with xts ones. I chose to create xts object in the first place because I also have time in my data (For example this is one instance of the DateTime column "2012-08-25 06:00:00") so I created my object this way:

myXtsObj = as.xts(mydata$var1, order.by = mydata$DatetimeVar)

I keep receiving the same error when I want to use some functions such as:

ma(myXtsObj, order = 4, centre = T) 
ets(myXtsObj)

Both ets and ma functions are not working which basically I am using to detect trends and seasonality in the data.

My questions:

1) I am not sure what is the correct value for "order" argument for my daily data in "ma" function. Any help?

2) The error appearing when I am using the two function and also when I try to convert the xts object to ts object using as.ts is:

Error in if (frequency > 1 && abs(frequency - round(frequency)) < ts.eps) frequency <- round(frequency) : missing value where TRUE/FALSE needed

What could be causing this error?

Answer 1

As the help files state, both ma() and ets() require ts objects as the first argument, not xts objects.

If you want to estimate a trend, setting order=7 in ma() will remove the day-of-week seasonality. A better method would be to use stl() or forecast::mstl()