[EMAIL PROTECTED] (Damian Franco) wrote in message news:<[EMAIL PROTECTED]>... > I am currently working on a project that requires some statistic > knowledge which I have very little. I was wondering if somebody can > help me by guiding me on the right direction. I have several > statistical books but I am not sure where to start reading. > I am including a short description of this project. Any > help/guide/reference that you can give me will be deeply appreciated. > > Thank you very much > ------- > > I have a database where I keep the daily circulation quantity of a > small local newspaper by zip code. > > Ex. > > Zip code Circulation Date > 33131 56 1/1/2002 > 33131 45 1/2/2002 > 33131 53 1/3/2002 > 33131 51 1/4/2002 > 33131 48 1/4/2002 > . > . > . > > 33131 85 12/27/03 > 33131 70 12/28/03 > 33131 45 12/29/03 > 33131 31 12/30/03 > 33131 51 12/31/03 > 33131 105 1/1/04 > . > . > > 33131 21 1/15/04 > > > We want to be able to do projections based on this data. For example, > Lets say we would like to know the circulation quantity in April 7, > 2004 in zip code 33131. > > The problem is that there are factors that we need to take in > consideration: > > One factor is the day of the week (weekday). For example, Sunday's > circulation is usually higher than Tuesday and Friday is higher than > Monday. > > Another factor is the week of the year. For example, thanksgiving week > circulation is higher because people want all the coupons inside the > newspaper to go shopping. > > The next factor is the month of the year. For example, in August a lot > of people go on vacation therefore circulation is lower than January. > In November we have a lot of people that come down here to Miami from > up north because of the weather so circulation is a lot higher. > > Another factor is the day of the year. For example, Election Day or > when the president delivers the State of the Union address at the end > of January. > > What exactly do you think I should read to be able to get a close > estimate of our circulation in the future? > > Thanks again
We at AFS have been working on this problem for a number of years. One of our customers Anheuser-Busch is using AUTOBOX to predict by DAY by STORE by PRODUCT taking into account DAY-OF-THE-WEEK , PRICE , WEATHER , WEEK-OF-THE-YEAR , HOLIDAYS and other promotions. Holidays often have a lead effect , contemporaneous effect and lag effects reflecting demand patterns around the individual holiday. Similarly, Carreker Corporation uses AUTOBOX to predict daily demand for cash at atm machines around the world. I would be glad to demonstrate the capability of AUTOBOX and FreeFore ( http://www.autobox.com/freef.exe ) a FREEWARE version. Please see http://www.autobox.com and pursue the thread to PROSUCTS/AUTOBOX/CASE STUDIES . The solution is to efficiently combine regresion ( with correct lead/lag structures ) and ARIMA structure reflecting omitted stochastic series and the effect of Intervention Variables ( Pulses, Level Shifts , Seasonal Pulses and LOcal Time Trends ) reflecting omitted deterministic variables. This is called a Transfer Function. If I can help please give me a call at 215-675-0652 Regards Dave Reilly Automaic Forecasting Systems . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
