Beruflich Dokumente
Kultur Dokumente
Application
Anthony Davison
c _2006
A short course based on the book
Bootstrap Methods and their Application,
by A. C. Davison and D. V. Hinkley
c _Cambridge University Press, 1997
Anthony Davison: Bootstrap Methods and their Application, 1
Summary
Useful when
standard assumptions invalid (n small, data not normal,
. . .);
standard problem has non-standard twist;
complex problem has no (reliable) theory;
or (almost) anywhere else.
Aim to describe
basic ideas;
condence intervals;
tests;
some approaches for regression
Anthony Davison: Bootstrap Methods and their Application, 2
Motivation
Motivation
Basic notions
Condence intervals
Several samples
Variance estimation
Tests
Regression
Anthony Davison: Bootstrap Methods and their Application, 3
Motivation
AIDS data
jk
= exp(
j
+
k
).
k unobs
jk
= exp(
j
)
k unobs
exp(
k
).
1 2 3 4 5 6 14 of 1992
1988 1 31 80 16 9 3 2 8 6 174
2 26 99 27 9 8 11 3 3 211
3 31 95 35 13 18 4 6 3 224
4 36 77 20 26 11 3 8 2 205
1989 1 32 92 32 10 12 19 12 2 224
2 15 92 14 27 22 21 12 1 219
3 34 104 29 31 18 8 6 253
4 38 101 34 18 9 15 6 233
1990 1 31 124 47 24 11 15 8 281
2 32 132 36 10 9 7 6 245
3 49 107 51 17 15 8 9 260
4 44 153 41 16 11 6 5 285
1991 1 41 137 29 33 7 11 6 271
2 56 124 39 14 12 7 10 263
3 53 175 35 17 13 11 306
4 63 135 24 23 12 258
1992 1 71 161 48 25 310
2 95 178 39 318
3 76 181 273
4 67 133
Anthony Davison: Bootstrap Methods and their Application, 5
Motivation
AIDS data
estimator complex
data complex
model non-standard
Anthony Davison: Bootstrap Methods and their Application, 7
Basic notions
Motivation
Basic notions
Condence intervals
Several samples
Variance estimation
Tests
Regression
Anthony Davison: Bootstrap Methods and their Application, 8
Basic notions
Handedness data
Table: Data from a study of handedness; hand is an integer measure
of handedness, and dnan a genetic measure. Data due to Dr Gordon
Claridge, University of Oxford.
dnan hand dnan hand dnan hand dnan hand
1 13 1 11 28 1 21 29 2 31 31 1
2 18 1 12 28 2 22 29 1 32 31 2
3 20 3 13 28 1 23 29 1 33 33 6
4 21 1 14 28 4 24 30 1 34 33 1
5 21 1 15 28 1 25 30 1 35 34 1
6 24 1 16 28 1 26 30 2 36 41 4
7 24 1 17 29 1 27 30 1 37 44 8
8 27 1 18 29 1 28 31 1
9 28 1 19 29 1 29 31 1
10 28 2 20 29 2 30 31 1
Anthony Davison: Bootstrap Methods and their Application, 9
Basic notions
Handedness data
Figure: Scatter plot of handedness data. The numbers show the mul-
tiplicities of the observations.
15 20 25 30 35 40 45
1
2
3
4
5
6
7
8
dnan
h
a
n
d
1 1
1
2 2 1 5
2
1
5
2
3
1
4
1
1
1 1
1
1
Anthony Davison: Bootstrap Methods and their Application, 10
Basic notions
Handedness data
Estimator
for unknown parameter .
Handedness data
y = (dnan, hand)
F puts probability mass on subset of R
2
is correlation coecient
For r = 1, . . . , R:
generate random sample y
1
, . . . , y
n
iid
F;
compute
r
using y
1
, . . . , y
n
;
1
,
2
, . . . ,
Use
1
,
2
, . . . ,
R
to estimate sampling distribution of
Algorithm: For r = 1, . . . , R:
generate random sample y
1
, . . . , y
n
iid
F;
compute
r
using y
1
, . . . , y
n
;
1
,
2
, . . . ,
R
Anthony Davison: Bootstrap Methods and their Application, 17
Basic notions
Nonparametric bootstrap
Bootstrap (re)sample y
1
, . . . , y
n
iid
F, where
F is EDF of
y
1
, . . . , y
n
Repetitions will occur!
Compute bootstrap
using y
1
, . . . , y
n
.
= (dnan, hand)
1
, . . . ,
See picture
Bootstrap replicates
r
used to estimate properties of
.
Bias of
as estimator of is
(F) = E(
[ y
1
, . . . , y
n
iid
F) (F)
estimated by replacing unknown F by known estimate
F:
(
F) = E(
[ y
1
, . . . , y
n
iid
F) (
F)
F) b =
= R
1
R
r=1
[ F) by
v =
1
R 1
R
r=1
_
_
2
Estimate quantiles of
by taking empirical quantiles of
1
, . . . ,
.
Histogram of t
t*
D
e
n
s
i
t
y
0.5 0.0 0.5 1.0
0
.
0
0
.
5
1
.
0
1
.
5
2
.
0
2
.
5
4 2 0 2 4
0
.
5
0
.
0
0
.
5
Quantiles of Standard Normal
t
*
Anthony Davison: Bootstrap Methods and their Application, 22
Basic notions
How many bootstraps?
2
0
2
4
R=199
Theoretical Quantiles
Z
*
4 2 0 2 4
2
0
2
4
R=999
Theoretical Quantiles
Z
*
Anthony Davison: Bootstrap Methods and their Application, 23
Basic notions
Key points
Estimator is algorithm
applied to original data y
1
, . . . , y
n
gives original
1
, . . . , y
n
gives
can be of (almost) any complexity
for more sophisticated ideas (later) to work,
must often be
smooth function of data
F F heroic assumption
F ,= F) reduce by thought
simulation (R ,= ) reduce by taking R large (enough)
Anthony Davison: Bootstrap Methods and their Application, 24
Condence intervals
Motivation
Basic notions
Condence intervals
Several samples
Variance estimation
Tests
Regression
Anthony Davison: Bootstrap Methods and their Application, 25
Condence intervals
Normal condence intervals
If
approximately normal, then
.
N( +, ), where
1/2
,
where (z
) = .
Replace , by estimates:
(F)
.
= (
F)
.
= b =
(F)
.
= (
F)
.
= v = (R 1)
1
r
(
)
2
,
giving (1 2) interval
b z
v
1/2
.
):
Transformed correlation coefficient
D
e
n
s
i
t
y
0.5 0.0 0.5 1.0 1.5
0
.
0
0
.
5
1
.
0
1
.
5
4 2 0 2 4
0
.
5
0
.
0
0
.
5
1
.
0
1
.
5
Quantiles of Standard Normal
T
r
a
n
s
f
o
r
m
e
d
c
o
r
r
e
l
a
t
i
o
n
c
o
e
f
f
i
c
i
e
n
t
Anthony Davison: Bootstrap Methods and their Application, 27
Condence intervals
Normal condence intervals
= (
) =
1
2
log(1 +
)/(1
)
for which compute
b
= R
1
R
r=1
r
, v
=
1
R 1
R
r=1
_
_
2
,
1
_
z
1
v
1/2
_
,
1
_
v
1/2
Hope properties of
1
, . . . ,
R
mimic eect of sampling from
original model.
Canonical example: Y
1
, . . . , Y
n
iid
N(,
2
). Then
Z =
Y
(S
2
/n)
1/2
t
n1
,
for all ,
2
so independent of the underlying
distribution, provided this is normal
V
1/2
If the quantiles z
of Z known, then
Pr (z
Z z
1
) = Pr
_
z
V
1/2
z
1
_
= 1 2
(z
V
1/2
z
1
V
1/2
z
_
= 1 2
so (1 2) condence interval is (
V
1/2
z
1
,
V
1/2
z
)
Anthony Davison: Bootstrap Methods and their Application, 30
Condence intervals
, V
) and hence
Z
V
1/2
R bootstrap copies of (
, V ):
(
1
, V
1
), (
2
, V
2
), . . . , (
R
, V
R
)
and corresponding
z
1
=
V
1/2
1
, z
2
=
V
1/2
2
, . . . , z
R
=
V
1/2
R
.
Use z
1
, . . . , z
R
to estimate distribution of Z for example,
order statistics z
(1)
< < z
(R)
used to estimate quantiles
V
1/2
z
((1)(R+1))
,
V
1/2
z
((R+1))
Anthony Davison: Bootstrap Methods and their Application, 31
Condence intervals
Why Studentize?
Studentize, so Z
D
N(0, 1) as n . Edgeworth series:
Pr(Z z [ F) = (z) +n
1/2
a(z)(z) +O(n
1
);
a() even quadratic polynomial.
is
Pr(Z
z [
F) = (z) +n
1/2
a(z)(z) +O
p
(n
1
).
Typically a(z) = a(z) +O
p
(n
1/2
), so
Pr(Z
z [
F) Pr(Z z [ F) = O
p
(n
1
).
Anthony Davison: Bootstrap Methods and their Application, 32
Condence intervals
If dont studentize, Z = (
)
D
N(0, ). Then
Pr(Z z [ F) =
_
z
1/2
_
+n
1/2
a
_
z
1/2
_
_
z
1/2
_
+O(n
1
)
and
Pr(Z
z [
F) =
_
z
1/2
_
+n
1/2
a
_
z
1/2
_
_
z
1/2
_
+O
p
(n
1
).
Typically = +O
p
(n
1/2
), giving
Pr(Z
z [
F) Pr(Z z [ F) = O
p
(n
1/2
).
Simpler approaches:
Basic bootstrap interval: treat
as pivot, get
((R+1)(1))
),
(
((R+1))
).
Percentile interval: use empirical quantiles of
1
, . . . ,
R
:
((R+1))
,
((R+1)(1))
.
((R+1)
)
,
((R+1)(1
))
,
where
= 0.0485,
Motivation
Basic notions
Condence intervals
Several samples
Variance estimation
Tests
Regression
Anthony Davison: Bootstrap Methods and their Application, 39
Several samples
Gravity data
Table: Measurements of the acceleration due to gravity, g, given as
deviations from 980,000 10
3
cms
2
, in units of cms
2
10
3
.
Series
1 2 3 4 5 6 7 8
76 87 105 95 76 78 82 84
82 95 83 90 76 78 79 86
83 98 76 76 78 78 81 85
54 100 75 76 79 86 79 82
35 109 51 87 72 87 77 77
46 109 76 79 68 81 79 76
87 100 93 77 75 73 79 77
68 81 75 71 78 67 78 80
75 62 75 79 83
68 82 82 81
67 83 76 78
73 78
64 78
Anthony Davison: Bootstrap Methods and their Application, 40
Several samples
Gravity data
Figure: Gravity series boxplots, showing a reduction in variance, a shift
in location, and possible outliers.
4
0
6
0
8
0
1
0
0
g
1 2 3 4 5 6 7 8
series
Anthony Davison: Bootstrap Methods and their Application, 41
Several samples
Gravity data
8
i=1
y
i
n
i
/s
2
i
8
i=1
n
i
/s
2
i
, V =
_
8
i=1
n
i
/s
2
i
_
1
,
giving
= 78.54, V = 0.59
2
and 95% condence interval of
1.96V
1/2
= (77.5, 79.8)
Anthony Davison: Bootstrap Methods and their Application, 42
Several samples
Gravity data: Bootstrap
, V
V
1/2
,
whose distribution is approximated by simulations
z
1
=
V
1/2
1
, . . . , z
R
=
V
1/2
R
,
giving
(
V
1/2
z
((R+1)(1))
,
V
1/2
z
((R+1))
)
(25)
, z
(975)
, giving interval (77.1, 80.3).
Anthony Davison: Bootstrap Methods and their Application, 43
Several samples
Figure: Summary plots for 999 nonparametric bootstrap simulations.
Top: normal probability plots of t
and z
= (t
t)/v
1/2
. Line on
the top left has intercept t and slope v
1/2
, line on the top right has
intercept zero and unit slope. Bottom: the smallest t
r
also has the
smallest v
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
. .
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. . .
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.. .
.
.
.
.
.
. .
.
.. .
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
..
.
.
.
.
..
.
.
. .
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
. .
.
.
.
.
.
.
..
.
. .
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
..
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
. .
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
..
.
.
.
..
.
.
.
.
. .
.
. .
.
.
.
.
.
.
.
.
. .
.
.
.
.
..
. .
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
Quantiles of Standard Normal
t
*
-2 0 2
7
7
7
8
7
9
8
0
8
1
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
. .
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
..
..
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
. .
.
.
. .
.. .
.
.
.
. .
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
. .
. .
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
. .
.
.
.
.
..
.
.
.
.
.
.
.
.
..
.
.
.
.
..
. .
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
..
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
..
.
.
. .
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
. .
.
. . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
. . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.. .
.
.
.
.
. ..
.
. .
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
. .
. .
.
.
. .
.
.
.
. .
.
. .
.
.
.
.
.
. .
.
.
.
.
.
..
.
.
. .
.
.
. .
.
.
.
. .
.
.
.
.
.
.
.
.
. .
.
.
.
..
. ..
.
..
. .
.
.
.
.
.
..
.
.
.
.
. .
.
.
.
.
.
.
. .
.
.
. .
.
.
.
.
.
. . .
.
.
.
.
.
.
.
.
.
. .
..
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
. .
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
. .
.
.
.
. .
.
.
.
.
.
.
. .. . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
...
.
.
.
.
.
.
.
..
.
.
.
.
.
.
. .
.
.
. .
.
.
.
.
.
.
.
. .
. .
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
..
.
. .
. .
.
.
..
.
.
.
..
. .
.
.
.
.
. .
.
.
. .
.
.
.
.
. .
.
.
.
.
.
...
.
.
.
. .
. .
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
. .
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
. .
.
.
.
.
. .
.
.
.
.
.
.
. .
.
.
.
..
. . .
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
..
.
.
.
. .
.
. .
.
.
.
. .
.
.
.
.
..
.
. .
.
.
.
.
.
.
. .
..
.
.
.
.
..
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
Quantiles of Standard Normal
z
*
-2 0 2
-
1
5
-
1
0
-
5
0
5
.
.
.
.
.
.
. . .
.
.
.
. .
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
. .
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
. .
.
.
.
.
. .
.
.
. .
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
. . .
.
.
.
.
.
. .
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
. .
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
. .
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
t*
s
q
r
t
(
v
*
)
77 78 79 80 81
0
.
1
0
.
2
0
.
3
0
.
4
0
.
5
0
.
6
0
.
7
.
.
.
.
.
.
. . .
.
.
.
. .
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
. .
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
..
.
.
.
.
. .
.
.
. .
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.. .
.
.
.
.
.
. .
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
..
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
. .
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
z*
s
q
r
t
(
v
*
)
-15 -10 -5 0 5
0
.
1
0
.
2
0
.
3
0
.
4
0
.
5
0
.
6
0
.
7
Anthony Davison: Bootstrap Methods and their Application, 44
Several samples
Key points
Motivation
Basic notions
Condence intervals
Several samples
Variance estimation
Tests
Regression
Anthony Davison: Bootstrap Methods and their Application, 46
Variance estimation
Variance estimation
Bootstrap sample y
1
, . . . , y
n
and corresponding estimate
1
, . . . , y
n
from
y
1
, . . . , y
n
, giving corresponding bootstrap estimates
1
, . . . ,
Q
,
1
, . . . ,
Suppose
= g(
.
N(,
2
/n)
Then provided g
() ,= 0, have (D2)
E(
) = g() +O(n
1
)
var(
) =
2
g
()
2
/n +O(n
3/2
)
Then var(
)
.
=
2
g
)
2
/n = V
Example (D3):
= Y ,
= log
)
.
= S()
2
/n, nd
transformation h such that varh(
)
.
=constant
= g(
1
, . . . ,
d
)
Anthony Davison: Bootstrap Methods and their Application, 49
Variance estimation
Anthony Davison: Bootstrap Methods and their Application, 50
Variance estimation
Nonparametric delta method
General approximation:
V
.
= V
L
=
1
n
2
n
j=1
L(Y
j
; F)
2
.
,
where H
y
puts unit mass at y. Close link to robustness.
l
2
j
,
usually obtained by analytic/numerical dierentation.
Anthony Davison: Bootstrap Methods and their Application, 51
Variance estimation
Computation of l
j
Write
in weighted form, dierentiate with respect to
Sample average:
= y =
1
n
y
j
=
w
j
y
j
w
j
1/n
Change weights:
w
j
+ (1 )
1
n
, w
i
(1 )
1
n
, i ,= j
so (D5)
y y
= y
j
+ (1 )y = (y
j
y) +y,
giving l
j
= y
j
y and v
L
=
1
n
2
(y
j
y)
2
=
n1
n
n
1
s
2
Interpretation: l
j
is standardized change in y when increase
mass on y
j
Anthony Davison: Bootstrap Methods and their Application, 52
Variance estimation
Nonparametric delta method: Ratio
= t(
F) =
_
xd
F(u, x)/
_
ud
F(u, x) = x/u
u
j
)/u,
giving
v
L
=
1
n
2
_
x
j
u
j
u
_
2
Anthony Davison: Bootstrap Methods and their Application, 53
Variance estimation
Handedness data: Correlation coecient
x
j
u
j
etc.:
=
xu xu
_
(x
2
x
2
)(u
2
u
2
)
_
1/2
,
from which empirical inuence values l
j
can be derived
Get
v
L
= 0.029
for comparison with v = 0.043 obtained by bootstrapping.
v
L
typically underestimates var(
) as here!
Anthony Davison: Bootstrap Methods and their Application, 54
Variance estimation
Delta methods: Comments
w
j
x
j
, w
j
1/n for x
and vary weights successively for j = 1, . . . , n, setting
w
j
= w
i
+, i ,= j,
w
i
= 1
for = 1/(100n) and using the denition as derivative
Anthony Davison: Bootstrap Methods and their Application, 55
Variance estimation
Jackknife
j
),
where
j
is value of
computed from sample
y
1
, . . . , y
j1
, y
j+1
, . . . , y
n
l
jack,j
, v
jack
=
1
n(n 1)
_
l
2
jack,j
nb
2
jack
_
Requires n + 1 calculations of
Motivation
Basic notions
Condence intervals
Several samples
Variance estimation
Tests
Regression
Anthony Davison: Bootstrap Methods and their Application, 58
Tests
Ingredients
P-value, p
obs
= Pr(T t
obs
[ M
0
) measures evidence
against M
0
small p
obs
indicates evidence against M
0
.
Diculties:
p
obs
may depend upon nuisance parameters, those of M
0
;
p
obs
often hard to calculate.
Anthony Davison: Bootstrap Methods and their Application, 59
Tests
Examples
Estimate p
obs
by simulation from tted null hypothesis
model
M
0
.
Algorithm: for r = 1, . . . , R:
simulate data set y
1
, . . . , y
n
from
M
0
;
calculate test statistic t
r
from y
1
, . . . , y
n
.
r
t
obs
1 +R
of
p
obs
= Pr(T t
obs
[
M
0
).
Take T =
(correlation coecient), with t
obs
= 0.509; this
is large in case of positive association (one-sided test)
F
1
(dnan
1
, . . . , dnan
n
) and from
F
2
(hand
1
, . . . , hand
n
),
then put them together to get bootstrap data
(dnan
1
, hand
1
), . . . , (dnan
n
, hand
n
).
, so
p =
1 + 18
1 + 9999
= 0.0019 :
hand and dnan seem to be positively associated
[: gives p = 0.004.
Anthony Davison: Bootstrap Methods and their Application, 62
Tests
Handedness data: Bootstrap from
M
0
15 20 25 30 35 40 45
1
2
3
4
5
6
7
8
dnan
h
a
n
d
3 10
1
3 2
2
4
2
1
1
1
2 1
2
1
1
Test statistic
P
r
o
b
a
b
i
l
i
t
y
d
e
n
s
i
t
y
0.6 0.2 0.2 0.6
0
.
0
0
.
5
1
.
0
1
.
5
2
.
0
2
.
5
Anthony Davison: Bootstrap Methods and their Application, 63
Tests
Choice of R
r
t
obs
1 +R
_
.
=
1
R
2
Rp
obs
(1 p
obs
) =
p
obs
(1 p
obs
)
R
so if p
obs
.
= 0.05 need R 1900 for 10% relative error (D7)
Handedness data:
0
= 0 , 1
0.95
, but
0
= 0 1
0.99
, so
estimated signicance level 0.01 < p < 0.05: weaker
evidence than before
Extends to tests of =
0
against other alternatives:
if
0
, 1
1
= (,
)/V
1/2
as
statistic to test =
0
0
)/V
1/2
Signicance level is
Pr
_
V
1/2
z
obs
[ M
0
_
= Pr(Z z
obs
[ M
0
)
= Pr(Z z
obs
[ F)
.
= Pr(Z z
obs
[
F)
Compare observed z
obs
with simulated distribution of
Z
= (
)/V
1/2
, without needing to construct null
hypothesis model
M
0
0
V
1/2
=
0.509 0
0.170
= 2.99
r
z
obs
1 +R
=
1 + 215
1 + 9999
= 0.0216
Test statistic
P
r
o
b
a
b
i
l
i
t
y
d
e
n
s
i
t
y
6 4 2 0 2 4 6
0
.
0
0
.
1
0
.
2
0
.
3
Anthony Davison: Bootstrap Methods and their Application, 67
Tests
Exact tests
Data Y
1
, . . . , Y
n
iid
Pois(), with unknown
(Y
j
Y )
2
/Y
.
2
n1
;
observed value is t
obs
= 55.15
Y
j
:
p
obs
= Pr(T t
obs
[
M
0
, W = w),
independent of , owing to suciency of W
Y
j
= 107.
Anthony Davison: Bootstrap Methods and their Application, 69
Tests
Example: Fir data
Figure: Simulation results for dispersion test. Left panel: R = 999 val-
ues of the dispersion statistic t
Take T =
(correlation coecient) again
), . . . , (dnan
n
, hand
n
)
where (1
, . . . , n
i,j
(y
ij
y
ij
)
2
y
ij
, y
ij
=
y
i
y
j
y
Get Pr(
2
24
38.53) = 0.048, but is T
.
2
24
?
Anthony Davison: Bootstrap Methods and their Application, 73
Tests
Exact tests: Contingency table
get p
.
= 0.08
Anthony Davison: Bootstrap Methods and their Application, 74
Tests
Key points
Motivation
Basic notions
Condence intervals
Several samples
Variance estimation
Tests
Regression
Anthony Davison: Bootstrap Methods and their Application, 76
Regression
Linear regression
Independent data (x
1
, y
1
), . . ., (x
n
, y
n
) with
y
j
= x
T
j
+
j
,
j
(0,
2
)
(1 h
j
)
1/2
.
(0,
2
)
) =
2
(X
T
X)
1
if model y = X + correct
Anthony Davison: Bootstrap Methods and their Application, 77
Regression
Linear regression: Resampling schemes
Model-based resampling:
y
j
= x
T
j
+
j
,
j
EDF(e
1
e, . . . , e
n
e)
Fixes design but not robust to model failure
Assumes
j
sampled from population
Case resampling:
(x
j
, y
j
)
EDF(x
1
, y
1
), . . . , (x
n
, y
n
)
Varying design X but robust
Assumes (x
j
, y
j
) sampled from population
Usually design variation no problem; can prove awkward in
designed experiments and when design sensitive.
Anthony Davison: Bootstrap Methods and their Application, 78
Regression
Cement data
Table: Cement data: y is the heat (calories per gram of cement) evolved
while samples of cement set. The covariates are percentages by weight
of four constituents, tricalciumaluminate x
1
, tricalcium silicate x
2
, tet-
racalcium alumino ferrite x
3
and dicalcium silicate x
4
.
x
1
x
2
x
3
x
4
y
1 7 26 6 60 78.5
2 1 29 15 52 74.3
3 11 56 8 20 104.3
4 11 31 8 47 87.6
5 7 52 6 33 95.9
6 11 55 9 22 109.2
7 3 71 17 6 102.7
8 1 31 22 44 72.5
9 2 54 18 22 93.1
10 21 47 4 26 115.9
11 1 40 23 34 83.8
12 11 66 9 12 113.3
13 10 68 8 12 109.4
Anthony Davison: Bootstrap Methods and their Application, 79
Regression
Cement data
Covariates compositional: x
1
+ +x
4
.
= 100% so X
almost collinear smallest eigenvalue of X
T
X is
l
5
= 0.0012
Plot of
1
against smallest eigenvalue of X
T
X
reveals
that var
1
) strongly variable
5
?
Anthony Davison: Bootstrap Methods and their Application, 80
Regression
Cement data
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
. .
.
.
.
. .
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
. .
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
. . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
..
.
.
.
.
.
.
.
. .
. .
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
. .
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
. .
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
. .
.
.
.
. . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
. . .
.
.
.
.
.
.
.
.
.
. .
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
. .
.
.
.
.
.
. .
.
.
.
.
.
.
.
. .
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
. ..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
. .
.
.
. .
.
.
.
.
.
.
. .
.
.
.
.
.
. .
.
.
. .
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
. .
.
.
.
.
.
.
.
. .
.
.
. .
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
. .
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
. . .
.
. .
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
. . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
smallest eigenvalue
b
e
t
a
1
h
a
t
*
1 5 10 50 500
-
1
0
-
5
0
5
1
0
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. . . . .
.
.
.
.
.
.
. .
.
.
.
.
.
.
. .
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
. .
. .
.
.
.
.
.
.
. .
..
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
. .
.
. . . .
.
.
.
.
.
..
.
.
.
.
.
.
.
. . .
.
. .
.
. .
.
. .
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
. ..
.
.
.
.
.
.
. .
.
.
.
.
.
..
.
. .
.
.
.
.
.
.
.
. . .
.
.
.
.
.
.
.
.
.
.
.
.
..
. .
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
. .
.
. .
.
.
.
.
.
.
.
.
.
. .
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
. .
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
. .
. .
.
..
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
. .
. . .
.
.
.
.
..
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
..
.
.
. .
.
.
. .
.
.
.
.
.
.
.
. .
. .
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
. .
.
.
.
.
.
.
.
.
.
. . .
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
. .
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. . .
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
. .
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. . .
.
.
.
.
.
.
. .
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
. . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
. .
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
. .
.
.
.
.
. .
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
. . .
.
.
. .
. .
..
.
.
. .
.
.
.
. .
.
.
.
. .
.
.
.
.
.
.
.
. . .
.
. .
.
.
.
.
. .
.
.
.
. . .
.
.
.
.
.
.
. . .
.
.
.
.
.
.
.
. . .
.
.
..
.
.
.
.
smallest eigenvalue
b
e
t
a
2
h
a
t
*
1 5 10 50 500
-
1
0
-
5
0
5
1
0
Anthony Davison: Bootstrap Methods and their Application, 81
Regression
Cement data
Table: Standard errors of linear regression coecients for cement data.
Theoretical and error resampling assume homoscedasticity. Resampling
results use R = 999 samples, but last two rows are based only on those
samples with the middle 500 and the largest 800 values of
1
.
2
Normal theory 70.1 0.74 0.72
Model-based resampling, R = 999 66.3 0.70 0.69
Case resampling, all R = 999 108.5 1.13 1.12
Case resampling, largest 800 67.3 0.77 0.69
Anthony Davison: Bootstrap Methods and their Application, 82
Regression
Survival data
dose x 117.5 235.0 470.0 705.0 940.0 1410
survival % y 44.000 16.000 4.000 0.500 0.110 0.700
55.000 13.000 1.960 0.320 0.015 0.006
6.120 0.019
Linear model:
log(survival) =
0
+
1
dose
dose
s
u
r
v
i
v
a
l
%
200 600 1000 1400
0
1
0
2
0
3
0
4
0
5
0
dose
l
o
g
s
u
r
v
i
v
a
l
%
200 600 1000 1400
-
4
-
2
0
2
4
Anthony Davison: Bootstrap Methods and their Application, 83
Regression
Survival data
Case resampling
Anthony Davison: Bootstrap Methods and their Application, 84
Regression
Generalized linear model
MLE
, tted values
j
, Pearson residuals
r
Pj
=
y
j
j
V (
j
)(1 h
j
)
1/2
.
(0, ).
Bootstrapped responses
y
j
=
j
+V (
j
)
1/2
j
where
j
EDF(r
P1
r
P
, . . . , r
Pn
r
P
). However
possible that y
j
, 0, 1, 2, . . . ,
r
Pj
not exchangeable, so may need stratied resampling
Anthony Davison: Bootstrap Methods and their Application, 85
Regression
AIDS data
jk
= exp(
j
+
k
)
Pearson residuals:
r
jk
=
Y
jk
jk
jk
(1 h
jk
)
1/2
Model-based simulation:
Y
jk
=
jk
+
1/2
jk
jk
Anthony Davison: Bootstrap Methods and their Application, 86
Regression
Diagnosis Reporting-delay interval (quarters): Total
period reports
to end
Year Quarter 0
1 2 3 4 5 6 14 of 1992
1988 1 31 80 16 9 3 2 8 6 174
2 26 99 27 9 8 11 3 3 211
3 31 95 35 13 18 4 6 3 224
4 36 77 20 26 11 3 8 2 205
1989 1 32 92 32 10 12 19 12 2 224
2 15 92 14 27 22 21 12 1 219
3 34 104 29 31 18 8 6 253
4 38 101 34 18 9 15 6 233
1990 1 31 124 47 24 11 15 8 281
2 32 132 36 10 9 7 6 245
3 49 107 51 17 15 8 9 260
4 44 153 41 16 11 6 5 285
1991 1 41 137 29 33 7 11 6 271
2 56 124 39 14 12 7 10 263
3 53 175 35 17 13 11 306
4 63 135 24 23 12 258
1992 1 71 161 48 25 310
2 95 178 39 318
3 76 181 273
4 67 133
Anthony Davison: Bootstrap Methods and their Application, 87
Regression
AIDS data
jk
;
estimate parameters from incomplete y
jk
get estimated row totals and truth
+,j
= e
b
k unobs
e
b
k
, y
+,j
=
k unobs
y
jk
.
Prediction error
y
+,j
+,j
1/2
+,j
studentized so more nearly pivotal.
Resampling schemes:
parametric simulation, tted Poisson model
parametric simulation, tted negative binomial model
nonparametric resampling of r
P
stratied nonparametric resampling of r
P
jk
= exp (j) +
k
,
where (j) is locally-tted smooth.
Useful when
standard assumptions invalid (n small, data not normal,
. . .);
standard problem has non-standard twist;
complex problem has no (reliable) theory;
or (almost) anywhere else.
Have described
basic ideas
condence intervals
tests
some approaches for regression
Anthony Davison: Bootstrap Methods and their Application, 94
Regression
Books