336.0
349.9
12.15
324.8

375.U
t;'Y'U:.~R::R:t:
evaluation
statistics
Hean
&aucered
Error
;wnr
:~tial1
Squared
Error
'·I~
Z;'uunlute
Error
Mean.
Percentage
Error
:~,tlr.
iilim:::>lute
Percentage
Error
"ian
>lr3!D0rtion,
UM
Rea~esBicon
proportion,
UR
Disturban.ce
proportion,
UD
17.05
313.98
17.719
17.05
5.0916
5.0916
4.933
0.92591
0.06982
0.0042748
,h,j
f\.'f~itst
;·'·.:",,:I·~
..
JtIH·.J
,'.
"
,
,
\
\
,
'~V~J\~\/
I
f
The Central Limit Theorem says that even for a properly specified model, I out
of20
draws will fall outside
of
the
95%
confidence
intervalthat
is
a 5% miss rate.
However, the plot above shows that
lout
of
5 values falls outside the
95%
confidence interval, with a miss rate
of
20%. Therefore,
this is not a good/reasonable model because the miss rate is too high.
Furthermore, this
is
not good forecasting because the forecasts are all higher
than the true values, so this model does not have good predicting power.
Problem 2
4.42089  10<0.261812)
=
0.4267
[2.87812
+
(100)0.084153  (20)(0.327667)]1/2
This has the student's t distribution with
23
d.f.
('nk
(312)
=
t'296
(0.05/2)
=
2.069
Since
Ie
<
t',
we fail to reject the null at the 5% level
of
significance and say that the effect
of
eating meat is not statistically different from onetenth
the effect
of
eating edible fats on coronary heart disease.
v/
Problem 3
ehd,
=
~I
+
~2edfat,
+
~Jmeat,
+
~4eig,
+
~5beer,
+
~6winet
+
u,
chd,
=
~I
+
~2edfat.
+
~Jmeat.
+
~4eig,
+
(~5

~6)beer,
+
~6wine,
+
~6beer,
+
u,
ehd,
=
~l
+
~2edfat.
+
~Jmeat.
+
~4cig,
+
obeer,
+
~6wine,
+
~6beer,
+
u,
ehd,
=
~I
+
~2edfat.
+
~Jmeat,
+
~4cigt
+
obeer,
+
~6(wine,
+
beer,)
+
u,
ehd,
=
~I
+
~2edfat,
+
~Jmeatt
+
~4cig,
+
obeer,
+
~6ZI,
+
u,
where ZI,
=
wine,
+
beer,
Hodel
2:
OLS,
using
observations
19471975
(T
29)
Dependent
variable:
chd