I am fitting an ARIMA model on a daily time series. Data are collected daily from 02-01-2010 to 30-07-2011 and are about newspaper sales. Since a weekly pattern in sales can be found (the daily average amount of copies sold is usually the same from Monday to Friday, then increases on Saturday and Sunday), I am trying to capture this "seasonality". Given the sales data "data", I create the time series as follows:


and then I use the auto.arima(.) function to select the best ARIMA model via AIC criterion. The result is always a non-seasonal ARIMA model, but if I try some SARIMAs model with the following syntax as example:

sarima1<-arima(salests, order = c(2,1,2), seasonal = list(order = c(1, 0, 1), period = 7))

I can obtain better results. Is there anything wrongs in the ts command / arima specification? The weekly pattern is very strong so I would not expect so many difficulties in capturing it. Any help would be very useful. Thank you, Giulia Deppieri


I have already changed some arguments. More precisely, the procedure selects ARIMA(4,1,3) as the best model when I set D=7, but AIC and the others good of fit indexes and forecasts as well) do not improve at all. I guess there's some mistakes due to confusion between seasonality and periodicity..?!

Auto.arima call used and output obtained:

modArima<-auto.arima(salests,D=7,max.P = 5, max.Q = 5)

 ARIMA(2,1,2) with drift         : 1e+20
 ARIMA(0,1,0) with drift         : 5265.543
 ARIMA(1,1,0) with drift         : 5182.772
 ARIMA(0,1,1) with drift         : 1e+20
 ARIMA(2,1,0) with drift         : 5137.279
 ARIMA(2,1,1) with drift         : 1e+20
 ARIMA(3,1,1) with drift         : 1e+20
 ARIMA(2,1,0)                    : 5135.382
 ARIMA(1,1,0)                    : 5180.817
 ARIMA(3,1,0)                    : 5117.714
 ARIMA(3,1,1)                    : 1e+20
 ARIMA(4,1,1)                    : 5045.236
 ARIMA(4,1,1) with drift         : 5040.53
 ARIMA(5,1,1) with drift         : 1e+20
 ARIMA(4,1,0) with drift         : 5112.614
 ARIMA(4,1,2) with drift         : 4953.417
 ARIMA(5,1,3) with drift         : 1e+20
 ARIMA(4,1,2)                    : 4960.516
 ARIMA(3,1,2) with drift         : 1e+20
 ARIMA(5,1,2) with drift         : 1e+20
 ARIMA(4,1,3) with drift         : 4868.669
 ARIMA(5,1,4) with drift         : 1e+20
 ARIMA(4,1,3)                    : 4870.92
 ARIMA(3,1,3) with drift         : 1e+20
 ARIMA(4,1,4) with drift         : 4874.095

 Best model: ARIMA(4,1,3) with drift        

So I assume the arima function should be used as:

bestOrder <- cbind(modArima$arma[1],modArima$arma[5],modArima$arma[2])
sarima1<-arima(salests, order = c(4,1,3))

with no seasonal component parameters and period specifications. Data and exploratory analysis show that the same weekly pattern can be approximatively considered for each week, with the only exception of August 2010 (when a consistent increase in sales is registered). Unfortunately I have no expertise in timeseries modeling at all, in fact I am trying this approach in order to find an alternative solution to other parametric e non-parametric models I have tried to fit for these problematic data. I have also many dependent numeric variables but they have shown low power in explaining the response variable: undoubtedly, the most difficult part to model is the time component. Moreover, the construction of dummy variables to represent months and weekdays turned out not to be a robust solution.

  • $\begingroup$ To determine the order of seasonality in my ts (daily data along 3years) I've used this code in Matlab: s=0;min=1e+07;n=length(x); for i=1:400 diff= x(i+1:n)-x(1:n-i); s(i)=sum(abs(diff)); if (s(i)<min) min=s(i);i end end It gives me 365 which is logical. $\endgroup$
    – nkabouche
    Commented Feb 21, 2018 at 10:19

2 Answers 2


If there is weekly seasonality, set the seasonal period to 7.

salests <- ts(data,start=2010,frequency=7) 
modArima <- auto.arima(salests)

Note that the selection of seasonal differencing was not very good in auto.arima() until very recently. If you are using v2.xx of the forecast package, set D=1 in the call to auto.arima() to force seasonal differencing. If you are using v3.xx of the forecast package, the automatic selection of D works much better (using an OCSB test instead of a CH test).

Don't try to compare the AIC for models with different levels of differencing. They are not directly comparable. You can only reliably compare the AIC with models having the same orders of differencing.

You don't need to re-fit the model after calling auto.arima(). It will return an Arima object, just as if you had called arima() with the selected model order.

  • $\begingroup$ thanks for pointing out my stupid mistake. I'll retract my answer. $\endgroup$
    – mpiktas
    Commented Aug 25, 2011 at 6:10
  • 1
    $\begingroup$ Than you so much for your very helpful suggestions. I am using the 2.19 version of the forecast package so I followed your advice and set the D parameter equal to 1 in the auto.arima() call. Now the best model selected for salests series is an ARIMA(1,0,0) with non-zero mean. Should I expect the specification of the seasonality part for the best model returned, I mean values of P,D,Q, or at least for D? $\endgroup$
    – Giulia
    Commented Aug 25, 2011 at 8:30
  • 2
    $\begingroup$ As long as your data has a frequency other than 1, seasonal ARIMA models will be considered. If a non-seasonal model is being returned, then either the seasonality is very weak or the data are not in a ts object with frequency > 1. $\endgroup$ Commented Aug 25, 2011 at 12:27

The problem with fitting seasonal ARIMA to daily data is that the "seasonal component" may only operate on the weekends or maybe just the weekdays thus overall there is a non-significnat "seasonal component". Now what you have to do is to augment your data set with 6 dummies representing the days of the week and perhaps monthly indicators to represent annual effects. Now consider incorporating events such as holidays and include any lead, contemoraneous or lag effect around these known variables. No there may be unusual values (pulses) or level shifts or local time trends in the data. Furthermore the day-of-the-week effects may have changed over time e.g. there was no Saturday effect for the first 20 weeks but a Saturday effect for the last 50 weeks.If you wish to post tour daily data I will give it a try and maybe other readers of the list might also contribute their analysis to help guide you through this.

  • $\begingroup$ In that case (IrishStat), wouldn't that be Mixed Modeling Technique instead of ARIMA. Lags are not taken anywhere in ARIMA, except for Box Jlung Test. Auto.arima (recent) fixes everything including Scaling of data, seasonality fluctuations (that's why I finds the best p,d,q parameters). $\endgroup$
    – wackyanil
    Commented Nov 18, 2015 at 5:39
  • $\begingroup$ It is called a Transfer Function and reflects a synergistic approach see autobox.com/pdfs/capable.pd starting with slide 42 . Auto.arima may work for simple cases but is not general enough in my opinion. If you have a data set in mind , make a new question and include it. $\endgroup$
    – IrishStat
    Commented Nov 18, 2015 at 13:19
  • $\begingroup$ @IrishStat did you mean ARIMA with intervention? day of week being dummy variables? and similar dummy variables for holidays? $\endgroup$ Commented Jan 1, 2016 at 6:27
  • $\begingroup$ Yes ..that would be my approach to daily data $\endgroup$
    – IrishStat
    Commented Jan 1, 2016 at 17:08

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.