Time series study the evolution of a variable being observed over time. The main features of a series before putting it to a forecasting process, summarized in four: 1. Cualitativas: I.e. given its importance and characteristics, they have gone through a previous process of qualitative analysis and is has been managing to identify that they are susceptible to predict its behaviour in the future. 2 Temporary: Companies time series are usually expressed in terms of the calendar, i.e.

values that are predicted are assigned to specific dates (months, days, hours, etc., depending on the basis of time working each company) 3.Quantitative: Corresponds to the value in the variables that take to express the different variations. 4 Probabilistic: i.e. the reliability of the forecast value occurring in the planned horizon. Time series classification mainly depends on the volume of data that have thus: type A: Series are high-volume. They are pretty regular so that the methods statistics like those used by Forecast Pro work well. Usually high-volume series are important for companies, and the consequences of errors in forecasting can be significant. So if they are not many, should be checked carefully one by one, and even make adjustments if deemed suitable by experience. Series type B: are volume medium.

Normally these series can be predicted with good accuracy by the methods of Forecast Pro. Since this group of items is not so crucial to the outcome of the company, it lends itself to predict them automatically. Type C: series are low-volume, and often represent more than 50% of the total of the series. Many of these series contain zeros with occasional sales 0 and eventually a big sale. The percentage of error in forecasts of the type C series is almost always very large but the consequences of this error are usually small. When forecasts type C systems appeared, they were almost never pronosticados. A prognosis is used by default (0 or) (1).