Demand Forecasting

ALL FORMULAS Y -
-
f ( Ni Nz
, ,
. . .

,
Nn )t E

Quality Regression model : Yi =

Both X , it Ei
Process capability measure Ccp ) =
( USL -

L SL ) 160 Estimated Regression Function : Ii bot bi Xii

-
-

' '

RSS ( Yi Fi ) t bi Xii ))
N 't tht t N"
FI
=

I factor for x Bar chart × I

-
-

I UCL & LCL for I

-

i
of IT
.

Average
- "

= → =

n '

Standard Error , Se = ( Yi Fil -

R
.
=
Max [n , > Nz ,
. . .

, Nn ] -

min [n , , Nz ,
. - i

, an ] Average of R → I UCL & LCL for R =

Factor for UCL / LCL in R chart x I in-kT

I I ÷ Factor to Approximate 95% Prediction Interval : In ± He

'
Estimator of standard Deviation = =
Estimate Standarte Deviation
Yi bot binti tbz Mit
stampftp.eucl/LCL--pI3x Estimated multiple Regression Analysis b kN ki
-

: t
samples )
-

Attribute control Chart SD ( f : average % of defects

. . .

→
Estimated SD = over all

( -7¥) )
"
R'a
,

Adjusted R -
= I

Flow rate of units processed Flow Rate of defects @ resource

Yield of Resource successfully @ resource ,
II
= = -

II
Flow Rate Flow Rate
mean Forecast Error = ein mean Absolute Deviation =
ein
'

mean squared Error =

.IE ,
ein
News vendor model Dt it Dt Lt t Dt N
Moving Average
- -
-

Ft Den
- . -

: =

N
Actual Demand Expected Actual Demand CM ) =
Forecast x AIF Ratio
AH patio =

Forecast
S.D. of Actual Demand =
SD of AIF Ratios
. x forecast simple Exponential smoothing :

pl Demand EQ ) =P loutcome of standard normal is z or lower ) New Forecast -

ax latest demand th a) x prev. forecast -

(Q ) coefficient of Variation of Demand o÷U

-0M
Q
-
-

Mt 2 Xo =

z =

workers having to make

]
unnecessary
differing of Idarets them
movements are
-

, *
Coverage cost ) Cost salvage cu Cun derage cost )
.

Co = -
Value -

-
Price -
cost
(

Fla ) Distribution of demand =P ( Demand EQ

}
)
coff
-
-

Ffa ) =
critical Ratio
cu
Cox FCQ ) =
Cu X Cl FCQ ) )
-

Find critical Ratio Find inside table Convert 2 to order

qty
→ →
z

=)
Expected lost Sales > t .
Find 2 Statistic .
( l
-
-
¥) 2 Look up in Standard Normal loss Function Table LK )
. .

3. Evaluate lost Sales =

0×42 )

Expected sales = M -

Expected lost sales making production

visible &
problems stopping
production upon detection of defects
Expected left over Inventory
=
Q -

Expected sales .
.

Expected profit =
[ ( price cost ) x
Expected sales ] [Cost salvage
-
-

value ) x Expected leftover Inv .

I

Expected sales Expected Sales =p Expected loss Sales

Expected Fill Rate = = -

Expected Demand U M workers in Toshiba had notebooks at their work stations that the used to jot down improvement ideas .

Minimum In Stock
Probability → t Find a statistics that yields target probability HQ ) =
Plz)
-

.
. .

2. Convert the z statistics into order qty .

In stock probability =
FCQ ) Stock out probability = I -

FCQ )

Minimum Fill Rate constraint → t . Find Lu ) = (F) ( I -

Fill Rate )

2 .
Find 2 statistics that yields from lost sales in step 1 .

This study source was downloaded by3 100000859636319

Convert z into fromorder
CourseHero.com
qty on 12-24-2022 15:45:48 GMT -06:00
.
.
 What to do about defects seven sources of waste

I .
Detect
Early
1 Overproduction 2.
Waiting 3 .
Transport 4 .
Overprocessing 5 .

Inventory 6 Rework
.
7- Motion

why ? It detects detected early capacity can increase and costs can reduce
are ,
. SS : sort , straighten ,
shine ,
Standardise , sustain
It a problem is detected early the number of defects can be reduced
,
.
Three Steps towards a JIT : t . .
Achieve one unit at - -
-
a -
time flow 2 .
Produce at Rate of customer Demand

If a problem is detected
early ,
the cause can be identified more
easily .
3. Implement pull systems ( Kamban )
HOW ? Culture that advocates two forms of process control
encourages workers to bring attention to problems To implement pull system
a ,
TPS :
.

Built-in tests signal problems automatically and immediately I . In Kamban -

based pull ,
the upstream replenishes what demand has withdrawn from the downstream .

2 . Prevent 2 . Make to -
-
order refers to the release of work into a
system only when a customer order has been received

adds value added Benefits with MRP materials supply directly to known demand
Why ? Inspection : non costs e capital for the inspection step : is linked
-

.
.

utilisation of facilities
does not remove the costs of scrap or rework Lower stock levels,
higher stock turn over Better customer service Higher
-
.

perfect l defects slip through ) MRP schedules link tactical

is never .
master planning with operational scheduling Early warnings
removes responsibility can be used for short term -

planning Assigns priorities for job supplying

-

materials
can become a bottleneck . Problems with MRP :

Prevention equality at the source) typically costs less than the combination of inspection ,
internal & external failure costs .
Reduced flexibility to deal with changes
HOW '
Systems become very complex
Eliminate defects
by developing capable business processes through standardisation and continuous can
, improvement .

Needs alot of detailed & reliable information

can be expensive & time -

consuming to implement .
Assumes that lead times are constant & interdependent of the quantities ordered .
Involves alot of data manipulation
order in which materials that they made specified by at materials
Ignores the are
really made & assumes are in the order the bill .

Basic Approach to Demand understand the objectives of forecasting Integrate demand planning and forecasting Identify major factors that influence demand forecast
Forecasting
understand and
identify customer segments Determine the appropriate forecasting technique Establish performance and error measures for the forecast .

Forecasting Requirements Accurate with small errors unbiased -

so they do not always under or over -

estimate demand Responsive to changes in demand In time for its purpose

.

Not affected by the odd unusual figure cost effective Easy to understand Influencing -
Factors time covered in the future
Availability of historical data type of product .

Relevance of historical data to the future variability of demand Accuracy needed cost of errors Benefits expected from the forecasts Amount of money & time available
&

Moving Average vs .

Exponential smoothing similarities :

Stationary process One parameter Ca or N ) lag behind trend Differences : ES All past data MA less N periods ES : : : less data

News vendor Model Implementation steps inputs 2 Generate a demand model 3 Choose an objective 4 Choose
:
.
I .
Gather economic .
.
. a
quantity to order

Distribution function : tells the probability the outcome will be a particular value or smaller always start low & increase towards l .
- O .

Density function : tells the probability of a particular outcome occurring .

the cost of ordering unit than ordered had you known demand
what
Co overage cost one more
you would have
= .

Cu cost the cost of ordering one fewer unit than what you would have ordered had you known demand
underage
= .

Methods to improve unfortunate News vendor losses Influence demand during 3 Find reactive supply ( even if more expensive) to reduce
t Improve forecast
accuracy z season underage
.

. .

costs 4 .
Reduce overage costs : discount outlets , transshipment , re manufacturing , recycling
Brainstorming technique that helps structure the process of identifying underlying causes of an outcome .
←
Ishikawa .

This study source was downloaded by 100000859636319 from CourseHero.com on 12-24-2022 15:45:48 GMT -06:00
Powered by TCPDF (www.tcpdf.org)