Choice of variables

The choice of output variables is carried out in accordance with some requirements which performance increases the effectiveness of experiment. Usually, the technology and economic output variables is distinguished. For example, the performance characteristics and profit are the economic variables, but percentage of target product, yield, quality of product are the technology variables. Often all these characteristics are connected with each other.

The studied process can be characterised by several output variables. Usually, the experimenter tries to reduce its to minimum. There are the different techniques of variables reducing. So, it is possible to change the goal of study, to “break” the process by the some units, to determine the correlation ties between the variables and so on. If the experimenter failed to find the main variable then he is needed to decide the compromise task between several output variables.

The requirements for output variables can be formulated by following manner:

-it should be measured;

-it must be effective (this is connected with correct formulation of problem);

-it must be statistically efficient (to have as less as possible variance that allows to differ the runs).

To choice of factors, it needs to perform the next requirements:

-the factors should be controllable, for example, the experimenter should be sure in being change of mixture stream from 30 l/h to 40 l/h;

-the accuracy of measurement should be sufficiently high (at least, it is higher by the order than the accuracy of output variable);

-all factors are in the range over which each factor will be varied (in the area of experiment or in the area of interest). It is important to investigate all factors that may be of importance. This process is usually a combination of practical experience and theoretical understanding. When the objective is factors’ screening, the special methods of screening analysis can be applied.

The requirements for the factors and output variables together. Since we are seeking the connection between factors and output variables, it is reasonable to consider mutual requirements. They are as follows:

- there is the complete accordance between output variable and factor in order to create the mathematics model;

- the constraints of factors and its combination must be such as output variable would become in itself range.

Of course, the main condition to choose variables is not only that is considered above. In particular, it is possible to formulate some statistics requirements and so forth.

Дата добавления: 2014-01-13