Time series with zero-values

Question

I have a daily time series about the sales of a product, my series start from 01/01/2016 until 31/08/2017.

There is something important about this series, it does have zero-values each Sunday because this shop doesn't work on Sundays, and it will not work on Sundays in the future, so I the expected value for Sundays is zero.

Once I run an Arima model, I am not sure about this forecast, I think it is not working but I do not know why.

I really need help here !

This is my data

https://drive.google.com/file/d/0BzIf8XvzKOGWSm1ucUdYUVhfVGs/view?usp=sharing

This is my code

x <- read.table("Series.txt", header = TRUE)

attach(x)

library(zoo)

myzoo<-zoo(x, seq(from = as.Date("2016-01-01"), to = as.Date("2017-08-31"), by = "day"))

plot(myzoo)

fit<-auto.arima(myzoo)

p1=forecast(fit,30)

plot(p1)

And this is my forecast

enter image description here

As you can see, I feel I'm doing something wrong, but I just do not know what.

Thank you for your help !

Answer 1

Approach 1:

Just set the Sunday observations to NAs. You can then replace the Sunday forecasts with 0s.

Approach 2:

Remove the Sundays and define the week as 6-days in length.