Statistical Design of Experiments
Statistical Design of Experiments
Statistical Design of Experiments
Statistical Design of Experiments
Residuals
Residuals
When a model is used to predict a
When a model is used to predict a
response, the quality of the model is
response, the quality of the model is
assessed by how much the
assessed by how much the
prediction varies from the actual
prediction varies from the actual
observed response
observed response
This difference is called a residual
This difference is called a residual
A residual is the observed value 
A residual is the observed value 
the predicted value
the predicted value
From the Yield Example:
From the Yield Example:
observed yield @ 40° and 1 bar = 66
observed yield @ 40° and 1 bar = 66
predicted yield @ 40° and 1 bar = 68.5
predicted yield @ 40° and 1 bar = 68.5
residual is observed
residual is observed


predicted
predicted
= 66 
= 66 
68.5 = 2.5
68.5 = 2.5
Residuals are studied to determine
Residuals are studied to determine
if the model is appropriate
if the model is appropriate