Page 36 - 6686

F , F
where p T - calculated and table-valued Fisher’s
criterion value respectively.
Tabular Fisher’s criterion value for degrees of freedom
f аd = N-l, f o = N∙(m-1) and given significance level q is
determined from Table 4.3.
If the condition (4.21) is satisfied, then linear equation of
regression is considered accurate. Scattering of experimental
data of variable of condition relatively equation of regression
has the same order, as the variance caused by random changes
of object of research (error of experiment).
2 2
When calculating F p, it is assumed that S аd > S о , but in
reality it could be vice versa. Conclusion about accuracy of the
model could be made without examining the condition (4.21).
If condition (4.21) is not satisfied, i.e. linear model is
inaccurate, then decision to reduce the range of variation factors
and repeat of experiment can be made. It is advisable to
introduce new factors into a plan of experiment, from
eliminated factors.
Task 4.6
Check the correctness of model for conditions of
Task 4.5.
Solving:
According to previous examples, we define the function of
~
y
response u for each line, by obtained equation of regression
by the formula (4.18):
~ 1 = 0,48-0,096∙(+1)+0,242∙(+1) = 0,626;
y
~
y 2 = 0,48-0,096∙(-1)+0,242∙(+1) = 0,818;
~
y
3 = 0,48-0,096∙(+1)+0,242∙(-1) = 0,142;
~
y 4 = 0,48-0,096∙(-1)+0,242∙(-1) = 0,334.
36