Databases Reference
In-Depth Information
FIGURE 4.27
The autocomplete function shows all the arguments but gives few clues as to their use.
sample data set in this example, m
=
1.032 and b
=
0.292. A graph of the original and predicted
(or fitted) data sets is shown in Figure 4.28.
Directly below the graph is a table of the original X/Y data set plus the Y predicted (or fitted)
values for the data set using the aforementioned slope and intercept. The final column Y - Y
Predicted
are also referred to as the residuals. In this instance, if a residual is positive, the fitted point lies
below the actual data, and if a residual is negative, it lies above the actual data. The r
2
value (also
known as the coefficient of determination value or COD) is located in cell B5 if the results are
placed into a new Worksheet. The r
2
value is a measurement that describes just how well the curve
is fitted to the data, where a 1.0 is a perfect fit and a 0 is indicative of zero correlation. In this
instance, r
2
0.957, which is considered a fair but less than optimal fit to the data. In general,
when r
2
> 0.98, the fit is considered good, and when r
2
< 0.90, it may be indicative that no meaningful
correlation exists. The fit in this example, therefore, provides a useful but not necessarily precise
prediction of any particular data point.
=
Linear regression of sample dataset
11
10
9
8
7
6
5
4
3
2
1
0
Y
Predicted Y
0123456
X
7891011
FIGURE 4.28
Graph of original and predicted regression results.
