R-年和周的每日数据和时间序列

Question

Hi I am new to R and this time series forecasting.

I have a sample data of sales by day for past 3 years and I would like to use this data set to produce plot to find seasonality and pattern.

My daily data format is like eg..

Date, Sales  
2010-01-01, 5 
2010-01-03, 3 
2010-01-04, 2 
.. 
2011-12-01, 4
..
2014-11-01, 1

What I want to see is similar to below plot but by week and year using ts function. Also, due to leap year some year has 53 weeks and some 52 weeks, any idea how this taken into account when plotting ?

Playing with this ts function is not easy to me so it will be great if someone could help with this ..

Answer 1

You should start by creating a ts object. Check ?ts for the syntax, but assuming your data above were stored in `data', it's basically

tsData <- ts(data, start=c(2010,1), frequency=365)

where start refers to the (year, month) and frequency is the number of samples per year. Then you can use plot.ts() to plot the entire time series

plot.ts(tsData)

To extract seasonal patterns or trends, you can use the decompose() function.

decompose(tsData)

Answer 2

Here is an sample

x <- 1:10
y <- 11:20
plot(x, y)
lines(x,y)

On your data sales and date.

You can replace y with date and x with sales. If you still have issue, plz post it to me.