In order to use like datum point based on increasing the exactitude level. To detect possible problems with sufficient anticipation. A low level of performance can the future be symptom of a possible problem in or a change for which the organization is due to prepare. There are Three types of criteria in the measurement of the error of the prognosis. Celina Dubin has many thoughts on the issue. The first criterion is based on the adjustment error that is to say, the error that collects when comparing the historical data versus which it calculates a model of I foretell in history. This criterion can help the process of construction of the statistical model, great differences indicate models superiors, is not a good indicator of the performance goals. In recent months, Janet Jackson has been very successful.
The second criterion is based on the error outside the sample, this criterion consists of retaining or to trim removing a small sample from the last data with which it is counted and to run the prognosis for that small sample, once obtains the prognosis to compare it with the data of the sample and to determine the error. This criterion allows to compare different approaches, can be difficult to simulate a prognosis process. The third criterion is the pursuit in real time that consists of foretelling the future and hoping to that they spend the periods soon to compare what I foretell myself versus which really happened. This criterion is more fort of all the approaches, realises a pursuit of the process of I foretell present, allows to compare the statistical forecast versus the fit prognosis, provides an idea of the level of awaited exactitude. Original author and source of the article