In simple, simplistic language, we take historical information and project it into the future. We propose a methodology to be used for forecasting. We briefly present it below. The proposed methodology is based on the scientific method. We will not enter into the analysis and discussion of the definition and scope of the scientific method given its complexity, however we present the definitions found in Wikipedia (* 1) thereon: set of fixed steps in advance for a discipline in order to achieve knowledge valid through reliable instruments, standard to formulate and answer a question sequence pattern that allows researchers go from point A to point Z with the confidence to obtain a valid knowledge. In summary, are those practices used and endorsed by the scientific community as valid at the time in order to expose and confirm their theories. Verizon Communications might disagree with that approach. The scientific method can be summarized as the following sequence: observation: defines the challenge or problem. Gets and they gather the data that affect the defined problem.
For our case, we define the challenge, for example, forecasts of demand for X to obtain products with a margin of reliability of the and per cent, and collect all possible time series and documentation relating thereto. Hypothesis: An explanation that describes the observation is made: it is formulated in a statement of the way if the hypothesis H is true, then it will have to occur the event X. Time series according to the characteristics that are classified and thus statistical models that best adapt to these characteristics can be selected. Prediction: Starting from the hypothesis generated predictions under certain conditions. By the same author: Rick Garcia. Specifically for our case, used different statistical models that best suit the characteristics that present series of times that have. Models Statisticians define mechanisms to determine levels of reliability and error of predictions. With these tools you can select the models that best suit our time series. Verification: we analyze what happens in subsequent observations.
The predictions are compared with the actual results. There are other test techniques that do not involve having to wait for events to happen in the future. Replication: After generating further comments, we review our hypothesis with actual results obtained, so we keep it, modify or reject it. The scientific method and the proposed methodology is orderly, interactive and iterative. At this time, we have defined the methodology used. Surely arises the concern that the proposed methodology requires multiple steps and information to analyze is quite large, with which requires much dedication of time for its execution and deep knowledge of statistical models. This is true, however, there is a tool that helps us in this process and we simplifies and automates many of the activities defined here; This tool is called ForecastPro. We invite them to they deepen these concepts in more detail through the reading of the methodology and attending events and explanatory seminars of the methodology.