没有合适的资源?快使用搜索试试~ 我知道了~
首页SPM CART_Classification_Modeling
SPM CART_Classification_Modeling
3星 · 超过75%的资源 需积分: 14 35 下载量 200 浏览量
更新于2023-03-16
评论 3
收藏 5.08MB PDF 举报
利用eCognition Developer输出的地类样本的特征值Excel,可以将其导入到CART决策树中,同样可以实现分类特征的自动选择和特征阈值的自动确定,利用CART决策树自动构建得到一个具有分类顺序的二叉树;然后将这个二叉树,应用到eCognition Developer中,构建分类规则集。这里是Salford Predictive Modeler(简称SPM)中构建CART决策数的操作方法。
资源详情
资源评论
资源推荐
SP
M
U
s
e
rs G
u
i
de
C
l
ass
i
f
i
cat
i
on Mode
li
ng
i
n CART
T
h
is
gu
i
de p
r
o
vi
de
s
a deta
il
ed de
scri
pt
i
on
of
cl
a
ssi
f
ic
at
i
on
m
ode
li
ng
i
n
C
A
RT
.
T
i
t
le
:
Cl
a
ssi
f
ic
at
i
on
M
ode
li
ng
i
n
C
A
RT
S
ho
r
t
D
escri
pt
i
on:
T
h
is
gu
i
de p
r
o
vi
de
s
a deta
il
ed de
scri
pt
i
on of
cl
a
ssi
f
ic
at
i
on
m
ode
li
ng
i
n
C
A
RT
.
Long
D
escri
pt
i
on:
T
he
m
a
i
n pu
r
po
s
e of th
is
gu
i
de
is
to p
r
o
vi
de a deta
il
ed o
v
e
rvi
e
w
of
cl
a
ssi
f
ic
at
i
on
m
ode
li
ng
i
n
C
A
RT
.
W
e
will
add
r
e
ss
the fu
ll
s
et of opt
i
on
s
a
v
a
il
ab
l
e du
ri
ng the
m
ode
l
s
etup a
s
w
e
ll
a
s
gu
i
de
y
ou th
r
ough a
ll
a
v
a
il
ab
l
e output
r
epo
r
t
s
and d
is
p
l
a
ys
. A
sim
p
l
e data
s
et
c
o
mi
ng f
r
o
m
the b
i
o
m
ed
ic
a
l
app
lic
at
i
on f
i
e
l
d
will
be u
s
ed to
ill
u
s
t
r
ate a
ll
of the
k
e
y
c
on
c
ept
s
.
Sa
l
fo
r
d S
ys
te
ms'
C
A
RT
is
the on
ly
de
cisi
on
-
t
r
ee
sys
te
m
ba
s
ed on the o
ri
g
i
na
l
C
A
RT
c
ode de
v
e
l
oped b
y
w
o
rl
d
-r
eno
w
ned Stanfo
r
d
U
n
iv
e
rsi
t
y
and
U
n
iv
e
rsi
t
y
of
C
a
li
fo
r
n
i
a at Be
rk
e
l
e
y
s
tat
is
t
ici
an
s
B
r
e
im
an,
Fri
ed
m
an,
O
ls
hen and Stone.
K
ey
Wo
r
d
s
:
C
A
RT
,
Cl
a
ssi
f
ic
at
i
on and
R
eg
r
e
ssi
on
Tr
ee
s
, de
cisi
on t
r
ee
s
, p
r
ed
ic
t
iv
e
m
ode
ls
, data
mi
n
i
ng
Cl
a
ssi
f
ic
at
i
on
M
ode
li
ng
i
n
C
A
RT
2
C
op
yri
ght, 2016 Sa
l
fo
r
d S
ys
te
ms
Introduction
T
he
m
a
i
n pu
r
po
s
e of th
is
gu
i
de
is
to p
r
o
vi
de a deta
il
ed o
v
e
rvi
e
w
of
cl
a
ssi
f
ic
at
i
on
m
ode
li
ng
i
n
C
A
RT
.
W
e
will
add
r
e
ss
the fu
ll
s
et of opt
i
on
s
a
v
a
il
ab
l
e du
ri
ng the
m
ode
l
s
etup a
s
w
e
ll
a
s
gu
i
de
y
ou th
r
ough a
ll
a
v
a
il
ab
l
e output
r
epo
r
t
s
and d
is
p
l
a
ys
. A
sim
p
l
e data
s
et
c
o
mi
ng f
r
o
m
the b
i
o
m
ed
ic
a
l
app
lic
at
i
on f
i
e
l
d
will
be u
s
ed to
ill
u
s
t
r
ate a
ll
of the
k
e
y
c
on
c
ept
s
.
Setting up a Classification Model in CART
Mod
eli
ng
D
a
t
ase
t
W
e
s
ta
r
t b
y
w
a
lki
ng th
r
ough a
sim
p
l
e
cl
a
ssi
f
ic
at
i
on p
r
ob
l
e
m
ta
k
en f
r
o
m
the b
i
o
m
ed
ic
a
l
li
te
r
atu
r
e.
T
he top
ic
is
l
o
w
b
ir
th
w
e
i
ght of ne
w
bo
r
n
s
.
T
he ta
sk
is
to unde
rs
tand the p
rim
a
ry
fa
c
to
rs
l
ead
i
ng to a bab
y
be
i
ng
bo
r
n
si
gn
i
f
ic
ant
ly
unde
rw
e
i
ght.
T
he top
ic
is
c
on
si
de
r
ed
im
po
r
tant b
y
pub
lic
hea
l
th
r
e
s
ea
rc
he
rs
be
c
au
s
e
l
o
w
b
ir
th
w
e
i
ght bab
i
e
s
c
an
im
po
s
e
si
gn
i
f
ic
ant bu
r
den
s
and
c
o
s
t
s
on the hea
l
th
c
a
r
e
sys
te
m
. A
c
utoff of
2500 g
r
a
ms
is
t
y
p
ic
a
lly
u
s
ed to def
i
ne a
l
o
w
b
ir
th
w
e
i
ght bab
y
.
T
he fo
ll
o
wi
ng
v
a
ri
ab
l
e
s
a
r
e a
v
a
il
ab
l
e:
L
O
W
-
B
ir
th
w
e
i
ght
l
e
ss
than 2500 g
r
a
ms
(c
oded 1
i
f
<2500, 0 othe
rwis
e
)
.
A
GE
-
Mother’s age.
FT
V
-
N
u
m
be
r
of f
irs
t t
rim
e
s
te
r
ph
ysici
an
visi
t
s
.
HT
-
His
to
ry
of h
y
pe
r
ten
si
on
(c
oded 1
i
f p
r
e
s
ent, 0 othe
rwis
e
)
.
LWD
-
Lo
w
Mother’s w
e
i
ght at
l
a
s
t
m
en
s
t
r
ua
l
pe
ri
od
(c
oded 1
i
f <110 pound
s
, 0 othe
rwis
e
)
.
P
TD
-
O
cc
u
rr
en
c
e of p
r
e
-
te
rm
l
abo
r
(c
oded 1
i
f p
r
e
s
ent, 0 othe
rwis
e
)
.
RAC
E
-
Mother’s ethnicity (coded 1, 2 or 3).
S
M
O
K
E
-
S
m
o
ki
ng du
ri
ng p
r
egnan
cy
(c
oded 1
i
f
sm
o
k
ed, 0 othe
rwis
e
)
.
U
I
-
U
te
ri
ne
irri
tab
ili
t
y
(c
oded 1
i
f p
r
e
s
ent, 0 othe
rwis
e
)
.
A
s
y
ou
mi
ght gue
ss
w
e a
r
e go
i
ng to e
x
p
l
o
r
e the po
ssi
b
ili
t
y
that
c
ha
r
a
c
te
ris
t
ics
of the
m
othe
r
,
i
n
cl
ud
i
ng
demographics, health status, and the mother’s behavior, might influence the probability of a low birth
w
e
i
ght bab
y
.
Beg
i
n b
y
l
oo
ki
ng fo
r
the
H
OSLE
M
.
C
SV data f
il
e that
s
hou
l
d be
l
o
c
ated
i
n
y
ou
r
Sa
m
p
l
e
D
ata fo
l
de
r
. You
m
a
y
c
on
s
u
l
t the gene
ric
pa
r
t
s
of th
is
m
anua
l
fo
r
a deta
il
ed de
scri
pt
i
on on ho
w
to open data
s
et
s
fo
r
m
ode
li
ng and
r
e
l
ated
sim
p
l
e
s
tep
s
m
ent
i
oned be
l
o
w
:
Open the
H
OSLE
M
.
C
SV data
s
et
l
o
c
ated
i
n the Sa
m
p
l
e
D
ata fo
l
de
r
.
Cl
a
ssi
f
ic
at
i
on
M
ode
li
ng
i
n
C
A
RT
3
C
op
yri
ght, 2016 Sa
l
fo
r
d S
ys
te
ms
P
r
e
ss
the
[Model…]
button
i
n the a
c
t
ivi
t
y
wi
ndo
w
(
un
l
e
ss
the
wi
ndo
w
is
s
upp
r
e
ss
ed
)
.
You
s
hou
l
d no
w
ha
v
e the
Mod
el Se
tup
wi
ndo
w
opened. In
w
hat fo
ll
o
ws
,
w
e de
scri
be the pu
r
po
s
e of a
ll
i
nd
ivi
dua
l
tab
s
.
Mod
el
T
a
b
T
h
is
gene
ric
tab
is
u
s
ed to
s
et up ana
lysis
t
y
pe, a
s
w
e
ll
a
s
s
e
l
e
c
t ta
r
get and p
r
ed
ic
to
rs
:
M
a
k
e
s
u
r
e that the
An
alysis E
ng
i
n
e
s
e
l
e
c
t
i
on bo
x
ha
s
CART
.
M
a
k
e
s
u
r
e that the
T
ar
g
e
t
T
y
p
e
is
s
et to
C
lassi
f
ica
t
i
on
/
Log
is
t
ic
B
i
n
ary
.
C
hange the
S
o
r
t:
s
e
l
e
c
t
i
on bo
x
to
F
ile Or
d
er
.
Se
l
e
c
t LO
W
a
s
the ta
r
get
v
a
ri
ab
l
e.
Se
l
e
c
t AGE,
R
A
C
E, S
M
OKE,
HT
,
U
I,
FT
V, P
TD
, and L
WD
a
s
p
r
ed
ic
to
rs
.
Spe
ci
f
y
R
A
C
E,
U
I, and
FT
V a
s
c
atego
ric
a
l
p
r
ed
ic
to
rs
.
Spe
ci
f
y
AGE, S
M
OKE, and B
WT
a
s
au
xili
a
ry
v
a
ri
ab
l
e
s
(s
ee be
l
o
w)
.
Cl
a
ssi
f
ic
at
i
on
M
ode
li
ng
i
n
C
A
RT
4
C
op
yri
ght, 2016 Sa
l
fo
r
d S
ys
te
ms
T
ar
g
e
t
Co
l
u
m
n
T
h
is
is
w
he
r
e
y
ou
s
pe
ci
f
y
the ta
r
get
v
a
ri
ab
l
e fo
r
the ana
lysis
.
MODEL <variable>
Example> MODEL LOW
Model (target and predictors) reset: LOW
Pre
d
ic
to
r
Co
l
u
m
n
T
h
is
is
w
he
r
e
y
ou
s
pe
ci
f
y
the p
r
ed
ic
to
rs
to be u
s
ed.
KEEP <v ar i abl e>, <v ar i a b l e>, …
Example> KEEP AGE, RACE, SMOKE, HT, UI, FTV, PTD, LWD
✓
W
hen the
T
ar
g
e
t
T
y
p
e
is
s
et to
C
lassi
f
ica
t
i
on
/
Log
is
t
ic
B
i
n
ary
, the ta
r
get
v
a
ri
ab
l
e
will
be
auto
m
at
ic
a
lly
def
i
ned a
s
c
atego
ric
a
l
and appea
r
wi
th the
c
o
rr
e
s
pond
i
ng
c
he
ckm
a
rk
at
l
ate
r
i
n
v
o
c
at
i
on
s
of the
Mod
el Se
tup
. S
imil
a
rly
, the
R
e
g
ressi
on
r
ad
i
o button
will
auto
m
at
ic
a
lly
c
an
c
e
l
the
c
atego
ric
a
l
s
tatu
s
of the ta
r
get
v
a
ri
ab
l
e. In othe
r
w
o
r
d
s
, the
s
pe
ci
f
i
ed
T
ar
g
e
t
T
y
p
e
dete
rmi
ne
s
w
hethe
r
the ta
r
get
is
t
r
eated a
s
c
atego
ric
a
l
o
r
c
ont
i
nuou
s
.
Cl
a
ssi
f
ic
at
i
on
M
ode
li
ng
i
n
C
A
RT
5
C
op
yri
ght, 2016 Sa
l
fo
r
d S
ys
te
ms
C
a
t
e
go
rical
Co
l
u
m
n
T
h
is
is
w
he
r
e
y
ou
s
pe
ci
f
y
w
h
ic
h p
r
ed
ic
to
rs
a
r
e
c
atego
ric
a
l
(
no
mi
na
l
o
r
d
iscr
ete
)
.
CATEGORY <var i abl e>, <var i abl e>, …
Example> CATEGORY FTV, LOW, RACE, UI
C
A
RT
s
uppo
r
t
s
"
h
i
gh
-l
e
v
e
l
c
atego
ric
a
l
v
a
ri
ab
l
e
s"
th
r
ough
i
t
s
p
r
op
ri
eta
ry
a
l
go
ri
th
ms
that qu
ickly
dete
rmi
ne
effe
c
t
iv
e
s
p
li
t
s
i
n
s
p
i
te of the daunt
i
ng
c
o
m
b
i
nato
rics
of
m
an
y-v
a
l
ued p
r
ed
ic
to
rs
.
T
h
is
featu
r
e
is
i
n
cr
ea
si
ng
ly
im
po
r
tant
i
n the p
r
e
s
en
c
e of
c
ha
r
a
c
te
r
p
r
ed
ic
to
rs
,
w
h
ic
h
i
n
"r
ea
l
w
o
rl
d
"
data
s
et
s
often ha
v
e
hund
r
ed
s
o
r
e
v
en thou
s
and
s
of
l
e
v
e
ls
.
W
hen fo
rmi
ng a
c
atego
ric
a
l
s
p
li
tte
r
, t
r
ad
i
t
i
ona
l
C
A
RT
s
ea
rc
he
s
a
ll
po
ssi
b
l
e
c
o
m
b
i
nat
i
on
s
of
l
e
v
e
ls
, an app
r
oa
c
h
i
n
w
h
ic
h t
im
e
i
n
cr
ea
s
e
s
geo
m
et
ric
a
lly
wi
th the nu
m
be
r
of
l
e
v
e
ls
. In
c
ont
r
a
s
t,
C
A
RT's
h
i
gh
-l
e
v
e
l
c
atego
ric
a
l
a
l
go
ri
th
m
i
n
cr
ea
s
e
s
li
nea
rly
wi
th t
im
e,
y
et
yi
e
l
d
s
the
opt
im
a
l
s
p
li
t
i
n
m
o
s
t
si
tuat
i
on
s
.
C
ha
r
a
c
te
r
v
a
ri
ab
l
e
s
a
r
e
im
p
lici
t
ly
t
r
eated a
s
c
atego
ric
a
l
(
d
iscr
ete
)
,
s
o the
r
e
is
no need to
"
de
cl
a
r
e
"
the
m
c
atego
ric
a
l
.
T
he
r
e
is
no
i
nte
r
na
l
limi
t on the
l
ength of
c
ha
r
a
c
te
r
data
v
a
l
ue
s
(s
t
ri
ng
s)
. You a
r
e
limi
ted
i
n
th
is
r
e
s
pe
c
t on
ly
b
y
the data fo
rm
at
y
ou
c
hoo
s
e
(
e.g., SAS, te
x
t, E
xc
e
l
, et
c
.
)
.
✓
Character variables (marked by “$” at the end of variable name) will always be treated as categorical
and
c
annot be un
c
he
ck
ed.
✓
Occasionally columns stored in an Excel spreadsheet will be tagged as “Character” even though the
v
a
l
ue
s
i
n the
c
o
l
u
m
n a
r
e
i
ntended to be nu
m
e
ric
. If th
is
o
cc
u
rs
wi
th
y
ou
r
data,
r
efe
r
to the
R
EA
D
I
N
G
D
A
T
A
s
e
c
t
i
on to
r
e
m
ed
y
th
is
p
r
ob
l
e
m
.
D
epend
i
ng
w
hethe
r
a
v
a
ri
ab
l
e
is
de
cl
a
r
ed a
s
c
ont
i
nuou
s
o
r
c
atego
ric
a
l
,
C
A
RT
will
s
ea
rc
h fo
r
d
i
ffe
r
ent
t
y
pe
s
of
s
p
li
t
s
. Ea
c
h ta
k
e
s
on a un
i
que fo
rm
.
C
ont
i
nuou
s
s
p
li
t
s
will
a
lw
a
ys
u
s
e the fo
ll
o
wi
ng fo
rm
.
A
c
a
s
e goe
s
l
eft
i
f
[split-variable] <= [split-value]
A node
is
pa
r
t
i
t
i
oned
i
nto t
w
o
c
h
il
d
r
en
s
u
c
h that the
l
eft
c
h
il
d
r
e
c
e
iv
e
s
a
ll
the
c
a
s
e
s
wi
th the
l
o
w
e
r
v
a
l
ue
s
of the
[
s
p
li
t
-v
a
ri
ab
l
e].
C
atego
ric
a
l
s
p
li
t
s
will
a
lw
a
ys
u
s
e the fo
ll
o
wi
ng fo
rm
.
A
c
a
s
e goe
s
l
eft
i
f
[
s
p
li
t
-v
a
ri
ab
l
e]
=
[level_i OR …level_j OR … level_k]
In othe
r
w
o
r
d
s
,
w
e
sim
p
ly
lis
t the
v
a
l
ue
s
of the
s
p
li
tte
r
that go
l
eft
(
and a
ll
othe
r
v
a
l
ue
s
go
ri
ght
)
.
One
s
hou
l
d e
x
e
rcis
e
c
aut
i
on
w
hen de
cl
a
ri
ng
c
ont
i
nuou
s
v
a
ri
ab
l
e
s
a
s
c
atego
ric
a
l
be
c
au
s
e a
l
a
r
ge
nu
m
be
r
of d
is
t
i
n
c
t
l
e
v
e
ls
m
a
y
r
e
s
u
l
t
i
n
si
gn
i
f
ic
ant
i
n
cr
ea
s
e
s
i
n
r
unn
i
ng t
im
e
s
and
m
e
m
o
ry
c
on
s
u
m
pt
i
on.
An
y
c
atego
ric
a
l
p
r
ed
ic
to
r
wi
th a
l
a
r
ge nu
m
be
r
of
l
e
v
e
ls
c
an
cr
eate p
r
ob
l
e
ms
fo
r
the
m
ode
l
.
W
h
il
e
the
r
e
is
no ha
r
d and fa
s
t
r
u
l
e, on
c
e a
c
atego
ric
a
l
p
r
ed
ic
to
r
e
xc
eed
s
about 50
l
e
v
e
ls
the
r
e a
r
e
lik
e
ly
to
be
c
o
m
pe
lli
ng
r
ea
s
on
s
to t
ry
to
c
o
m
b
i
ne
l
e
v
e
ls
unt
il
i
t
m
eet
s
th
is
limi
t.
W
e
s
ho
w
ho
w
C
A
RT
c
an
c
on
v
en
i
ent
ly
do th
is
fo
r
y
ou
l
ate
r
i
n the
m
anua
l
(s
ee Int
r
odu
c
t
i
on to
D
ata B
i
nn
i
ng
)
.
剩余74页未读,继续阅读
o刘非刘o
- 粉丝: 3476
- 资源: 6
上传资源 快速赚钱
- 我的内容管理 收起
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
会员权益专享
最新资源
- ExcelVBA中的Range和Cells用法说明.pdf
- 基于单片机的电梯控制模型设计.doc
- 主成分分析和因子分析.pptx
- 共享笔记服务系统论文.doc
- 基于数据治理体系的数据中台实践分享.pptx
- 变压器的铭牌和额定值.pptx
- 计算机网络课程设计报告--用winsock设计Ping应用程序.doc
- 高电压技术课件:第03章 液体和固体介质的电气特性.pdf
- Oracle商务智能精华介绍.pptx
- 基于单片机的输液滴速控制系统设计文档.doc
- dw考试题 5套.pdf
- 学生档案管理系统详细设计说明书.doc
- 操作系统PPT课件.pptx
- 智慧路边停车管理系统方案.pptx
- 【企业内控系列】企业内部控制之人力资源管理控制(17页).doc
- 温度传感器分类与特点.pptx
资源上传下载、课程学习等过程中有任何疑问或建议,欢迎提出宝贵意见哦~我们会及时处理!
点击此处反馈
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功
评论2