Predicting Wining Percentage for the NFL

1. Descriptive Statistics

WinPct
TakeInt
TakeFum
GiveInt
GiveFum
DefYds/G
RushYds/G
PassYds/G
FGPct

Mean
0.50
15.84
12.13
15.84
12.16
315.94
112.46
203.45
80.77
Standard Error
0.04
0.95
0.53
0.88
0.58
4.43
4.17
6.64
1.20
Median
0.56
16.00
11.50
15.50
12.00
316.80
106.00
205.30
80.50
Mode
0.69
16.00
11.00
16.00
9.00
316.80

76.50
Standard Deviation
0.21
5.38
3.00
4.98
3.28
25.05
23.62
37.56
6.81
Sample Variance
0.04
28.91
9.02
24.78
10.78
627.34
557.72
1410.42
46.39
Kurtosis
-1.28
1.02
0.10
1.19
-0.45
1.41
-0.63
0.07
-0.37
Skewness
-0.05
0.45
0.61
0.58
0.19
0.83
0.49
-0.39
0.15
Range
0.75
26.00
12.00
24.00
13.00
113.40
88.00
158.70
28.90
Minimum
0.13
5.00
7.00
6.00
6.00
277.80
71.10
118.60
66.70
Maximum
0.88
31.00
19.00
30.00
19.00
391.20
159.10
277.30
95.60
Sum
16.01
507.00
388.00
507.00
389.00
10110.00
3598.70
6510.50
2584.70
Count
32.00
32.00
32.00
32.00
32.00
32.00
32.00
32.00
32.00

2. Multiple regression model

Coefficients
Standard Error t Stat
P-value
Intercept
0.976996
0.579844
1.68493
0.103528
DefYds/G
-0.00333
0.001291
-2.57907
0.015675
RushYds/G
0.004249
0.001353
3.140408
0.004061
PassYds/G
0.000735
0.000873
0.842015
0.407176
FGPct
-0.00064
0.004715
-0.13649
0.89245

The estimated regression model is
WinPct =0.976996-0.00333*DefYds/G+0.004249*RushYds/G+0.000735*PassYds/G-0.00064*FGPct

The rergewssion coefficients can be interpreted as
For a unit increase in DefYds/G, the WinPct decrease by 0.00333 units.
For a unit increase in RushfYds/G, the WinPct increase by 0.00429 units.
For a unit increase in PassYds/G, the WinPct increase by 0.000735 units.
For a unit increase in FGPct, the WinPct decrease by 0.00064 units.

The adequacy of the model is evaluated using the R2 value . Here R2 =0.4187

Thus 41.87% variability in WinPct can be explained