Assumptions
1.
Past
into
Error
measurements
2.
Quantified
form
of
data
Trend s
3.
continued
pattern
1.
Qualitative
:
xmathax
historical
data
.
a
biased
1.
Trend
①
¢
③
m
nemanh-ajts.tn#.nlAt-Ftty
improved
2.
Quantitative
:
2.
seasonal
e'
die
L
penalize
Erin
's
3.
Cyclical
table
①
Eh
④
her
rat
'
'
Y¥¥aEge
=
In
¥
.
,
lAt
large
error
→
Explanatory
-
relationships
Ex
.
Regression
analysis
4.
Random
disasters
At
→
time
series
-
historical
patterns
so
.pl?o,.mneasnEsauared=fIFAt-FtT
"
s
'D
'
*
StationaryMod
:
steady
constant
meant
random
variation
-
j
penance
n
n
around
'
n'
intr
models
Description
Formulas
Tren d
Analysis
rise
=
In
.§
,
lat
-
ft
)
'
n
.
,
an
-
e
-
simple
-1
flexible
Future
I
=
a
tbt
'
frnecdnatem
a
variance
adjusted
①
NAIVE
-
baseline
Etty
=
At
L
projected
value
<
actual
present
be
Eta
-
lEt)(E#
x
stationary
⑦
simple
iaiiiiiiiiaiiiiaiaiiaiiiiiiiim
"
.
Eat
.
"
sis
.int?a..:..::::i
:i÷:÷:÷÷÷÷:÷÷
:
:
Moving
AV
"
→
responsive
,
,
i
:O
a
=
A-
-
BE
÷i
.
Seasonal
Ty
-
a.
of
cycle
annual
)
'
factor
'
✓
easy
a
Stable
no
.
of
periods
L
factor
in
an
seasons
-
X
trend
seasonality
v
,
agg
trend
,
ignore
complex
used
to
forecast
E
'
Ent
A-
'
-
EA
behind
relationship
n
use
yea
tbf
n
-
I
d.
n'
in
int
www.NMNWW
:
↳
total
demand
forecast
in
yr
it
Is
:÷÷:÷±÷÷i÷÷÷±÷÷÷÷
:/
!
!
:÷÷÷÷÷÷÷÷÷:÷i÷÷÷::i
:÷÷÷÷÷:÷÷÷÷÷÷÷÷÷
:
"
:
:
:
"
"
"
"
'
iii.
sun
.
Smoothing
-
recent
data
>
old
1
(
previous
determined
'
forecast
-
-
lags
a
trend
Ifn
:
A
,
)
Smoothing
constant
'
nib
for
PRESENT
period
0
AM
X
stable
inn
trend
(
Trio
)
at
✓
Stable
Si
'
-
121=0.28
9:40
.tt#teigeeasrg;gn
⑤
Adjusted
✓
trend
expo
smoothed
51.9
trend
factor
predict
season
n
in
ft
:S
)
Exponential
+
seasonality
Aff
-
-
Ft
-11
+
Ttt
,
y
:
40.7
-14.4Gt
-
-
40.7+4.46157=63
Forecast
SMOOTHING
t
t
for
period
season
1
:
Snfs
=
0.281631
i
17.64
darth
-
d)
Fe
/
Bette
,
-
ft
)
-111
-
B)
Tt
①
Ho
benttereminden.fm
stable
)
②
Hq
Hypothesissteps
}
①
Goodness
of
fit
tests
1)
observed
③
Test
stats
-
Check
distribution
2)
Expected
④
critical
region
(
find
)
pi
is
given
:
Multinomial
/
Emstaiieiffff
⑤
conclusion
inborn
observed
/
k
outcomes
-
Normal
VS
Homogeneity
tests
\
Ii
:
!
!
!
;e'
as
:
'
:p
!
!
"pi
.
.fi?9iii9iydf=K-rU-XnOfindprobpoow:iPooanoior.uao=
'
to
:
indiscrete
ru
Binomial
Geo
-
metric
Bernoulli
't
(
Xi
)
=
Mpi
③
find
E
y
category
W
j
category
\
Pi
's
are
functions
iunieopmflinosmi.fi?yphergeoiinr
.
L
,
arraign
'
em
.
①
ioii
column
noii
columns
R
i
Enn
Aliunde
Formula
Pex
:#
a-
et
nigra
rrwoiosvr.mn#-1
pig
!
!
Caution
hp
;
75
if
not
,
you
column
①
Ho
:
pm
=
Pmi
Pr
,
=
pay
for
each
j
-
-
1943,4
Distribution
is
continuous
⑦
Ha
:
Ho
is
not
true
Kw
:
K2
p
n
Nog
=
14
i
0.25
normal
Eiinpio
area
"
!
'
estimate
ftp.r
.
,
=
E
(
Observed
-
E
)
'
③
Test
Statistics
Distribution
Ex
.
8
intervals
aniioo
,
E
'
100×18=125
Eas
-
D
.
#
XZ
=
10
df
:
3
*
E
25
,
if
not
merge
columns
!
n
'
as
→
Lor
more
-1
=
E
(
hi
-
hp
;)
'
0.07
Cp
-
value
C
0.025
⑦
contingency
Idf
:
re
-
no
-
i
)
i
:
'
Tpi
④
critical
Region
qnorosiwjwprob.li#
1)
Homogeneity
Ho
:
Pij
:
Rj
=
.
.
.
-
-
Pij
X
!
.gg
,
,
=
9.384
-
Voy
>
E'
o.osasn.ndgpyq.mg
ow
@
going
"g
for
each
jink
.
.
.aJ
:
.
Reject
at
95%
Confidence
(
I
fixed
mainly
population
divided
into
j
categories
E
:
row
total
x
Col
total
iiioaume
"
-
f
mann
.
routed
a
j
column
totals
are
random
IUVO
@
Nosed
)
n
Independence
ftp.iiiimgni
)
find
relationship
of
factors
2
factors
in
-
nails
independent
TWIN
Ho
:
Pij
=p
,
-
*
pj
for
i.
Ig
.
.
.gI
j
.
-
ra
.
.
.aJ
1
population
2
factors
frow
in
column
j
)
total
sample
fixed
a
row
-1
column
totals
are
random
Et
.
X
:
NO
-
Of
defects
,
Xv
poison
in
:
go
)
Hou
Poison
laid
Han
Does
not
follow
poisoner
)
space
:{
Salalah
)
y
'
:
13111.83
't
.
.
.
.
PIKA
:e°
:
0.368
22.08
0
!
Pirin
:
e-
'
In
'
'
-
0.368
=
7.23 6
i.
Rejection
Xhosa
4
-
o
-
i
pay
-
-
e
"
'
'
0.184
:
as
,
PIX
-
-
3)
-
e-}
'
-
0.061
:
Do
not
Reject
Ho
Prx
?
3)
↳i
black
Amarna
up
25
Ho
:
Pij
=
Pit
Pj
for
i
:...
-
I j
=
...
]
89787
-
L
X
eijing
p
!
Not
reject
Her
Mel
zo
e
