SC0x 2T2023 Module 1 Annotated

CTL.
SC1x -Supply- Supply

CTL.SC0X Chain & Logistics
Chain Fundamentals
Analytics
Introduction to Logistics & Supply

Chain Management
MIT Center for

Transportation & Logistics
"Bananas". Licensed under Creative Commons Attribution-Share Alike 2.5 via Wikimedia Commons -
http://commons.wikimedia.org/wiki/File:Bananas.jpg#mediaviewer/File:Bananas.jpg
"Red High Heel Pumps" by Almighty1 at en.wikipedia - Own workTransferred from en.wikipedia. Licensed under Public domain via Wikimedia Commons -
http://commons.wikimedia.org/wiki/File:Red_High_Heel_Pumps.jpg#mediaviewer/File:Red_High_Heel_Pumps.jpg
“Toyota IIA opened” by Hoikka1 Licensed under Creative Commons Attribution-Share Alike 3.0 Unported via Wikimedia commons –
http://commons.wikimedia.org/wiki/File:Toyota_IIA_opened_2.jpg
“Cement Bags” by StaraBlazkova Licensed under Creative Commons Attribution-Share Alike 3.0 Unported via Wikimedia commons
CTL.SC1x - Supply Chain and Logistics Fundamentals Lesson: Introduction to Supply Chains 2
http://commons.wikimedia.org/wiki/File:Cement_bags.jpg
Five Questions
1. What is a supply chain?

2. What are logistics and supply chain management?
3. How should you view SCM?
4. What are the challenges
5. Why should you care?
What is a Supply Chain?
Two or more parties linked by a flow of resources –
typically material, information, and money – that ultimately
fulfill a customer request.
Information (order)
Information (status)
Material (delivery)
Money (payment)
What is a supply chain?
Customer
Tire Retailer
Rubber
Mfg Supplier Hardware
Supplier
Gear
Supplier
Supplier
Wholesaler
Customer
Pedal
Supplier
Supplier
Bicycle Retailer
Casting
Smelter
Plant
Wholesaler Customer
Frame
Supplier
Retailer
Pigment
Supplier
Paint
Supplier Retailer Customer
Chemical
Mfg
• The primary purpose of a SC is to satisfy customer needs.

n Only one source of revenue – the customer
n Payments between parties are just fund exchanges
n Division of intra-SC payments are a function of power, market
conditions, etc.
• Supply Chains try to maximize the total value generated

n = [What customer pays] – [Total effort expended to fulfill]
Logistics versus Supply Chain Mgmt
What is Supply Chain Management?
Logistics involves . . . “managing the flow of items, information, cash
and ideas through the coordination of supply chain processes and
through the strategic addition of place, period and pattern values.”
MIT Center for Transportation & Logistics
“Supply Chain Management deals with the management of materials,

information, and financial flows in a network consisting of suppliers,
manufacturers, distributors, and customers.“
Stanford Supply Chain Forum
“Call it distribution or logistics or supply chain management. By

whatever name it is the sinuous, gritty, and cumbersome process by
which companies move material, parts, and products to customers.”
Fortune (1994)
Logistics vs. SCM
According to the Council of Supply Chain Management Professionals . . .
n Logistics management is that part of supply chain management that

plans, implements, and controls the efficient, effective forward and reverse
flow and storage of goods, services and related information between the
point of origin and the point of consumption in order to meet customers'
requirements.
n Supply chain management encompasses the planning and

management of all activities involved in sourcing and procurement,
conversion, and all logistics management activities. Importantly, it also
includes coordination and collaboration with channel partners, which can
be suppliers, intermediaries, third party service providers, and customers.
In essence, supply chain management integrates supply and demand
management within and across companies.
Source: Used by permission, Council of Supply Chain Management Professionals, http://cscmp.org/.
Supply Chain Management
Perspectives
Supply Chain Perspectives
Image Source: Arntzen, B. MIT Center for Transportation & Logistics, Hi-Viz Research Project (2013)
Supply Chain Perspectives
Customers
Manufacturers
Suppliers Distributors
Process View of Supply Chains
Customer
• Four Primary Cycles
n Customer Order Cycle Customer Order Cycle
n Replenishment cycle
Retailer
n Manufacturing Cycle
n Procurement Cycle Replenishment Cycle
• Cycles Occur Between Stages
n Interactions differ at each stage Distributor
• Not every SC will have all 4 Cycles
Manufacturing Cycle
Manufacturer
Procurement Cycle
Adapted from Chopra & Meindl “Supply Chain Management”
Supplier
Macro Process (Software) Perspective
Firm
Customer
Supplier
SRM ISCM CRM

“source” “make, move, store” “sell”
Supplier Relationship Internal Supply Chain Customer Relationship

Management Management Management
n Sourcing n Strategic Planning n Marketing
n Negotiation n Demand Planning n Selling
n Buying n Supply Planning n Call Centers
n Design and Supply n Fulfillment n Order Management
Collaboration
Different macro processes should be aligned

n Fall under different groups (Marketing-CRM, Manufacturing-
ISCM, Purchasing-SRM)
n Forecasts, schedules, design, etc. should be coordinated
Supply-Chain Operations Reference (SCOR) Model
Plan P1 Plan Supply Chain
P2 Plan Source P3 Plan Make P4 Plan Deliver P5 Plan Returns
Source Make Deliver

Suppliers
Customers
S1 Source Stocked Products M1 Make-to-Stock D1 Deliver Stocked Products
S2 Source MTO Products D2 Deliver MTO Products

M2 Make-to-Order
S3 Source ETO Products M3 Engineer-to-Order D3 Deliver ETO Products
Return Return
Source Deliver
Enable
Source: Supply Chain Council
Traditional Functional Perspective
• Purchasing / Procurement • Order Processing
n What to buy from who n Receiving, Entry & Status
n Corporate vs Group n Order Management
• Inventory Control • Transportation
n How much to stock where n Inbound versus Outbound
n Trigger points n Domestic versus International
n Replenishment plan n Modal control (Rail, TL, LTL,
• Warehousing Parcel, Air, etc.)
n Storage, Mixing, Break bulk • Customer Service
n Pick Pack and Ship n Geographic
n What to stock where in WH n Product Line Specific
• Materials Handling • Planning Group
n How to move product n Facility Location
n Packaging, containerization n Network Design
n Storage layout n Demand Planning
Systems Perspective
of Supply Chains
Supply Chain as a System
• Take an Engineering Systems Perspective
n What is a variable and what is a constraint?
n Continuous expansion of decision variables
n Increases potential for improvement but increases both
complexity and coordination requirements
Material Handling
Order Processing
Inventory Mgmt
Transportation
Customer Svc
Warehousing
Purchasing
Objective: Deliver at lowest transport cost

Variable:
wSelect carrier to tender each load to
Constraints:
wShip everything each day
wMust deliver within specified windows
Material Handling
Order Processing
Inventory Mgmt
Transportation
Customer Svc
Warehousing
Purchasing
Objective: Deliver at lowest transport cost

Variables:
Objective: Deliver at lowest total cost
Variables: wSelect carrier to tender each load to
wSelect carrier to tender each load to wSelect time windows to deliver
wSelect time windows to deliver Constraints:
wSelect when to ship what from where wShip everything each day
Constraints:
wDeliver within negotiated time frame
Material Handling
Order Processing
Inventory Mgmt
Product Design
Manufacturing
Transportation
Customer Svc
Warehousing
Purchasing
Objective:
wDesign, build, and deliver at lowest total cost
Variables:
wSelect time windows to deliver
wSelect when to ship what from where
wSelect where to stock which form of product
Constraints:
wDeliver within negotiated time frame
Objective:
Material Handling wMaximize on-shelf availability
Order Processing
Inventory Mgmt
Product Design
Manufacturing
Transportation
Customer Svc
Variables:
Warehousing
Purchasing
Supplier
Retailer
wSelect time windows to deliver
wSelect when to ship what from where
wSelect where to stock which form of
product
wSelect contract relationships
wSelect who should control replenishment
wWhich channel member should perform
which function
Why is this so hard to do? Constraints:
wTotal delivered cost to shelf
What are the major challenges?
(Why is it so hard?)
Why is it so difficult? 1/2
• Metrics – How do you measure a system?
n Trade-off of Breadth vs. Validity of metrics
n Outcome Based Logistics - Perfect order, Perfect Shelf
• Politics and Power of Players - Who wins?

n Mom & Pop Shop versus Mega-Stores
n Mega Retailers vs. Mega CPG Manufacturers
• Visibility – Who can see what and how quickly?

n Data are stored separately
n All parties do not have equal access to data
n Massive data ≠ Shared & accessible information
Why is it so difficult? 2/2
• Uncertainty - Who knows what is going to happen?
n Variable demand of product (shorter lifecycles)
n Variable manufacturing yield
n Unreliable sourcing of raw materials
n Inconsistent transit lead times
• Increased complexity – Why is it getting harder?

n Exploding number of Stock Keeping Units (SKUs)
n Higher and diverging customer demands
n New & merging channels (Omni-Channel)
• Global operations – Why don’t we ever close?

n Most firms source & sell across globe
n Multiple regions, time zones, languages, & cultures
Why do we care?
Why do we care?
• Supply Chains . . .
n Span the globe and cannot be managed as an isolated function,
n Have become critical to any organization’s operations, and
n Connect functions, divisions, and business units within a firm
as well as across firms.
Evolved into being both a Bridge and a Shock Absorber
By Wsvan; cropped by Beyond My Ken (talk) 01:29, 14 October 2013 (UTC) (Own work) [GFDL (http://www.gnu.org/copyleft/fdl.html) or CC-BY-SA-3.0 (
http://commons.wikimedia.org/wiki/File:Leonard_P._Zakim_Bunker_Hill_Bridge_-_Boston,_MA.jpg
"Mini Shocks" by Avsar Aras - Own work. Licensed under Creative Commons Attribution-Share Alike 3.0 via Wikimedia Commons - http://commons.wikimedia.org/wiki/
CTL.SC1x - Supply Chain and Logistics Fundamentals
File:Mini_Shocks.JPG#mediaviewer/File:Mini_Shocks.JPG Lesson: Introduction to Supply Chains 28
Example: Variability of #2 Diesel Pricing
4.8
4.6
4.4
4.2
On Highway Price of #2 Diesel ($/gallon)
3.8
3.6
3.4
3.2
2.8
2.6
2.4
2.2
2
Oct-06 Apr-07 Oct-07 Apr-08 Oct-08 Apr-09 Oct-09 Apr-10 Oct-10 Apr-11 Oct-11 Apr-12 Oct-12 Apr-13 Oct-13 May-14 Nov-14
Data Source: U.S. Energy Information Administration, http://www.eia.gov/petroleum/gasdiesel/diesel_proc-methods.cfm

Logistics Costs per GDP
GDP Inv Cost Trans Cost
Year ($T) ($B) ($B)
1981 3.21 259 228
1985 4.35 227 274
1990 5.98 283 351
1995 7.66 302 441
2000 10.29 385 594
16.0%
2005 13.1 410 739
2010 14.96 429 769
14.0% 2013 16.8 469 862
12.0%
Percentage of US GDP
10.0%
8.0%
6.0%
4.0%
2.0%
1980 1985 1990 1995 2000 2005 2010 2015
Year
US Inventory Holding Cost US Transportation Cost Total US Logistics Cost
Data Sources: Council of Supply Chain Management (2014) 25th Annual State of Logistics Report
Still, why do we care?
Functional Logistics Supply Chain
Key Skills Technically Narrow Coordination
Influence Local & Controllable Across entire Supply Chain
Leadership Hierarchical, Direct, Influential, Indirect, ‘Soft’

‘Hard’
Risk Management Robust or Reactionary Planned Response & Flexibility
Measurement Single Focus Multi-faceted
Technology Isolated Optimization Visibility & Coordination

Approach
Technology Self-Hosted, On-site Cloud or SaaS
Platform
Scope / Reach Regional / National Global / Multi-National
Supply Chain Management is a growing and evolving discipline.

CTL.SC1x -Supply Chain & Logistics Fundamentals
Questions, Comments, Suggestions?

Use the Discussion!
caplice@mit.edu
MIT Center for
Transportation & Logistics "Boxer (2)" by Dezidor - Own work. Licensed under Creative Commons Attribution 3.0 via
Wikimedia Commons - http://commons.wikimedia.org/wiki/
File:Boxer_(2).jpg#mediaviewer/File:Boxer_(2).jpg
Analytics Basics:
Models, Algebra, & Functions
MIT Center for

Transportation & Logistics ctl.mit.edu
Decision making is at the
core of supply chain management.
n How many facilities should I open and where?

n What transportation option should I use?
n How should I trade-off service and cost?
n Where should I source my raw material from?
n How should I share risk with my customers/suppliers?
n How much inventory should I have?
n What is my demand for next year?
n How can I make my supply chain more resilient?
Analytical models are used to
make supply chain decisions
2
Model Classification
Human is part of Human is external to

the modeling process the modeling process
Real Operational Analytical

Gaming Simulation
World Exercise Model
Increasing degree of
abstraction and speed
Increasing degree of
realism and cost
Source: Bradley, Hax, Magnanti 1977 3

Classification of Models
Strategy Evaluation Strategy Generation
Deterministic Simulation Linear Programming

Certainty Econometric Models Network Models
Systems of Simultaneous Integer and MILP
Equations Non-Linear Programming
Input-Output Models Control Theory
Monte-Carlo Simulation Decision Theory
Uncertainty Econometric Models Dynamic Programming
Stochastic Processes Inventory Theory
Queuing Theory Stochastic Programming
Reliability Theory Stochastic Control Theory
Source: Bradley, Hax, Magnanti 1977 4

Categories of Mathematical Models
Model Functional Independent OR/MS

Category Form f(.) Variables Techniques
Descriptive known, unknown or Simulation, PERT,
What has well-defined uncertain Queueing Theory,
happened? Inventory Models
Predictive unknown, known or under Regression Analysis,

What could ill-defined decision maker’s Time Series Analysis,
happen? control Discriminant Analysis
Prescriptive known, known or under Classic Opt., LP, MILP,

What should well-defined decision maker’s CPM, EOQ, NLP,
we do? control
5
Source: Ragsdale, 2004
Roadmap for the Course
• Deterministic – Prescriptive Modeling
n Basic functions & algebra
n Classical optimization (calculus)
n Math programming (LPs, IPs, MILPs, & Non-Linear)
• Stochastic/Uncertainty – Predictive & Descriptive
n Basic probability and distributions
n Statistical analysis (hypothesis testing)
n Econometric modeling (regression)
n Simulation
6
SCx Approach to Modeling
- Educating Drivers not Mechanics!
All images are CC0 Public Domain from https://pixabay.com and https://openclipart.org 7
Mathematical Functions
8
Mathematical Functions
“ . . . a relation between a set of inputs
and a set of permissible outputs with the
property that each input is related to
exactly one output.”
source: Wikipedia
we say:
“f of x” or that “y is a function of x”
y = f (x)
If given a value for x, then I can compute the value for y.
Example: f(x) = x2
x= 2 then y = f(2) = 22 = 4
x= 3.4 then y = f(3.4) = 3.42 = 11.56
x= -2 then y = f(-2) = (-2)2 = 4
Image by By Wvbailey Public Domain, https://commons.wikimedia.org/w/index.php?curid=10562739
9
Typically, constants are denoted by
Linear Functions
letters from the start of the alphabet
(a, b, c, . . .) while variables are letters
from the end of the alphabet (x, y, z).
“y changes linearly with x”
This is the
dependent function, f(x).
variable
y = ax + b constant,
fixed value, or
independent y-intercept
slope or variable
variable value
1
a
b
x
10
Examples: Linear Functions
• Truckload Transportation Costs:
cost = f(distance) = $200 + 1.35 $/km * (distance)
• Warehousing Costs
cost = f(# cases) = €2,500 + 2.5 €/case * (# cases)
• Profit Equation
profit = f(volume) = (r-c) * v - d
where:
r = revenue per item(¥/item)
c= cost per item (¥/item)
v = volume sold (items)
d = fixed cost (¥)
11
Quadratic Functions
12
Quadratic Functions y
a>0
Parabola - Polynomial function of degree 2

where a, b, and c are numbers and a≠0
This is the x
dependent
variable function, f(x).
2 a<0
y
y = ax + bx + c
x
independent constants
variable
• When a>0, the function is convex (or concave up)

• When a<0, the function is concave down
13
Finding Roots of Quadratic y
(iii)
(i)
• The root(s) of a quadratic (ii)
n Values of x for when y=0 x

n There can be 2, 1, or 0 roots
• Two methods for finding roots (i) y = 2x2
(ii) y = 2x2 - 6x +4
(iii) y = 3x2 - 4x + 2
n Factoring:
w Find r1 and r2 such that ax2+bx+c = a(x-r1)(x-r2)
n Quadratic equation
2
− b ± b − 4ac
r1 ,r2 =
2a
14
Example: Finding Roots
(i) y = 2x2 so that a=2, b=c=0
− b ± b 2 − 4ac
- 0 ± 0 2 - 4(2)(0) 0 r1 ,r2 =
r1 , r2 =
2(2)
= =0
4
2a
(ii) y = 2x2 – 6x + 4 so that a=2, b=-6, c=4
y
6 ± (−6)2 − 4(2)(4) 6 ± 36 − 32 6 ± 2
r1, r2 = = = (iii)
2(2) 4 4 (i)
(ii)
8 4
r1 = =2 r2 = =1
4 4
x
(iii) y = 3x2 – 4x + 2 so that a=3, b=-4, c=2
4 ± (−4)2 − 4(3)(2) 4 ± 16 − 24 4 ± −8 (i) y = 2x2

r1, r2 = = = (ii) y = 2x2 - 6x +4
2(3) 6 6 (iii) y = 3x2 - 4x + 2
r1, r2 are complex numbers
15
Quadratic Functions in Practice
16
• Example: Manufacturing iWidgets – what price to set?
n Cost of producing iWidgets is a linear y
function of the number produced, x: a 1
w cost =f(# made) = 500,000 + 75x b
n Demand for iWidgets is also a linear x
function of the price, p: y
w unit sales = f(price) = 20,000 – 80p

n So then: x
w Revenue = (20,000-80p)p = 20,000p – 80p2
w Costs = 500,000+75(20,000-80p) = 2,000,000 – 6000p
w Profit = Revenue – Costs = − b ± b 2 − 4ac
= 20,000p – 80p2 – (2,000,000 – 6,000p) r1 ,r2 =
= -80p2 + 26,000p – 2,000,000 2a
r1 = 125
What are the root(s) of this equation? r2= 200
17
iWidget Example Profit = -80p2 +26,000p – 2,000,000
Maximum profit
Price at which we
maximize profit
Roots of quadratic
18
• Example: Parcel Trucking – impact of weight
Parcel carriers combine many orders into a single shipment.
The cost of an individual order is a function of its weight, w.
However, it is not linear – it is tapering.
cost =f(weight) = 33 + 0.067w – 0.00005w2
€ 55
Cost Per Order (€) as a Function of Weight (kg)
€ 50
Cost Per Order (€)
€ 45
€ 40
€ 35
€ 30
0 50 100 150 200 250 300
Weight of Order (kgs)
19
Other Common Functional Forms
20
Power Function y=f(x) = axb
The shape of the curve is dictated by the value of b
3.00
if b<0, then function decreases
asymptotically to f(x)=0 and if b>1, then function
2.50
note that f(0) is undefined, increases exponentially
e.g., y=ax-1=a/x
2.00
1.50
1.00 if 0<b<1, then function

tapers away from linear
0.50
if b=1, then the
function is linear
0.00
- 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
y(b=1) y(b=1.5) y(b=.5) y(b=-1)
21
Exponential Functions y=abx
80.00
• Exponential functions have very fast growth

y=3x
70.00
• Widely used in physics, biology, economics,
60.00 finance, computer science, etc.
• Example: compound interest
50.00 Future = Principal (1 + Rate)Number_of_Periods
y=ex
= 100 (1 + 0.20)5 = 248.8
40.00
Euler’s Number, e, is a constant, like π
30.00 e = 2.71828 . . .
n
20.00 lim ⎛ 1 ⎞
e= ⎜1+ ⎟
n →∞ ⎝ n⎠
y=2x
10.00
0.00
- 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00
22
Logarithms
23
Other Inverse Relationships
Logarithms y=ax
y=a+x
çè
çè
x=y÷a
x=y-a
y=bx çè logb(y)=x
y is the value of b raised x is the power that I need to
to the xth power. raise the base, b, to equal y.
100 = 10x log10(100) = x x=2

5 = 10x log(5) = x x≅0.7
1 = ex loge(1) = ln(1)=x x=0
e = ex ln(e) = x x=1
Properties of Logarithms Examples:

• log(xy) = log(x) + log(y) • ln(3*5) = ln(3)+ln(5) = 2.71
• ln(12/7) = ln(12) – ln(7) = 0.54
• log(x/y) = log(x) – log(y) • ln(36) = 6 ln(3) = 6.59
• log(xa) = a log (x) • log(3*52) = log(3) + 2 log(5) = 1.88
Adapted from http://www.purplemath.com/modules/logs.htm 24

Practical Example: Doubling Time
• You have invested a sum of money that has an interest rate of 7% annually.
How many years, T, will it take to double in value?
n We know that F=P(1+r)n and we want to find the n where F=2P
n F=2P=P(1+r)n which reduces to 2=(1+r)n=(1.07)n
n We can transform this by taking the ln or log of both sides:

w ln(2) = n ln(1.07)
w Rearranging gives us: n = ln(2) / ln(1.07) = 0.693 / 0.067658 = 10.24 = T
w We could also use log10 where T = log(2)/log(1.07) = 10.24
n The investment will double in value in 10.24 years.
70
Number of Years to Double in

• Can we come up with a general 65
60
55
equation or approximation? 50
45
Value (T)
40
n We know that T =ln(2) / ln(1 + r) 35
30
n Plotting this for T=f(r) . . . 25
20
looks like T=ar-1=a/r 15
10
5
n Turns out T≈ 70/r -
- 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Annual Interest Rate (r)
Adapted from http://www.purplemath.com/modules/logs.htm 25

Multivariate Functions
26
Multivariate Functions
INPUT (x1, x2, . . . xn)
These are just functions with more than one
independent variable.
we still just say:

still just a “y is a function of x1, x2, . . . xn)”
single output
OUTPUT y = f(x1, x2, . . . xn)
y = f (x1, x2 ,...xn )
Example: f(x1, x2) = x1 + 2x2 + 5x1x2

x1= 2, x2 = 4 then y = f(2,4) = 2 + 2(4) + 5(2)(4) = 50
x1= -1, x2 = 0 then y = f(-1,0) = -1 + 2(0) + 5(-1)(0) = -1
x1= 0, x2 = -½ then y = f(0, -½) = 0 + 2(-½) + 5(0)(-½) = -1
Image by By Wvbailey Public Domain, https://commons.wikimedia.org/w/index.php?curid=10562739

27
Examples: Multivariate Functions
• Parcel Trucking – impact of weight & distance
Parcel carriers combine many orders into a single shipment.
The cost of an individual order is a function of its weight, w,
and the distance.
cost =f(weight, distance) = c1 + c2w + c3w2 + c4d + c5d2 + c6dw
• Total Logistics Cost Equation

cost = f(Demand, Order Cost, Order Size, ) = cD + AD/Q where:
D = annual demand (items)
c = cost per item (¥/item)
A = cost per order (¥/order)
Q = order size (items/order)
28
Properties of Functions
29
Properties of a Function: Convexity
A function is convex if it “holds water”
f (x) is convex if f [ λ x2 + (1− λ )x1 ] ≤ λ f (x2 ) + (1− λ ) f (x1 )
for all values of λ , where 0 < λ < 1, and for all pairs of x, where x1 < x2
• Example: f(x) = 3x2 - 4x + 2
• Pick x1 and x2: x1= -1 and x2= 2 λf(x2) + (1-λ)f(x1) = 6.21
• Find f(x1) and f(x2) y f(x2)=f(2) = 6.0

f(x1) = f(-1) =3(-1)2-4(-1)+2 = 6.3
f(x1)=f(-1) = 6.3
f(x2) = f(2) = 3(2)2-4(2)+2 = 6.0
• Pick a λ: 0.3 (so 1-λ = 0.7)
• Find Left Hand Side
• λx2 + (1-λ) x1 = (0.3)(2) + (0.7)(-1) = -0.1 f(λx2 + (1-λ)x1)
= f(-0.1) = 2.43
• f(λx2 + (1-λ) x1) = 3(-0.1)2-4(-0.1)+2 = 2.43
• Find Right Hand Side
• λf(x2) = (0.3)f(2) = (0.3)(6) = 1.80
• (1-λ)f(x1) = (0.7)f(1) = (0.7)(6.3) = 4.41
• λf(x2) + (1-λ)f(x1) = 1.8 + 4.41 = 6.21
x1=-1 x2=2 x
• Since 2.43 ≤ 6.21 the function is convex
30
Properties of a Function: Continuity
• A function f(x) is continuous at point c iff this means “if and only if”
1. f(c) is defined Does the function “work” at the selected x value?
2. limx->c f(x) exists Does the function “work” as we approach the selected x value?
3. limx->c f(x) = f(c) As we approach the selected x value, does the function return the
correct value?
• Some Notes:
n A function is continuous if you can draw it without lifting pen from paper!
n If the function has holes, jumps, or asymptotic conditions – it is not continuous
(1) f (x) = 3x 2 for all c values? Yes y
1 No, violates 1
(2) f (x) = @ c = 0?
x asymptotic at x=0
1
(3) f (x) = @ c > 1? Yes x
x
⎧⎪ 2 No, step at x=2,
x , x<2
(4) f (x) = ⎨ violates 3
⎪⎩ 3x, x ≥ 2
31
Key Points from Lesson
32
Key Points from Lesson (1/2)
• Different models used for different purposes
n Descriptive – what has happened?
n Predictive – what could happen?
n Prescriptive – what should we do?
• Functions y=f(x)
n Linear functions where y= ax + b
n Quadratic functions where y= ax2 + bx + c
n Power functions where y= axb
n Exponential functions where y= abx
33
Key Points from Lesson (2/2)
• Logarithms
n y=bx is equivalent to logb(y) = x
n Natural log ln(y) = loge(y)
• Multivariate functions y=f(x1, x2, . . . xn)
n Multiple inputs still lead to single output value
• Properties of functions
n Convexity – does the function “hold water”?
n Continuity – can I draw the function without
lifting my pencil
34
Questions, Comments, Suggestions?
Use the Discussion Forum!
“Wilson – realizing he is asymptotic to the door”

Yankee Golden Retriever Rescued Dog
(www.ygrr.org)
MIT Center for caplice@mit.edu
Introduc)on to Data
Management
MIT Center for

Informa)on Growth 1 Exabyte =
1 billion Gigabytes
• The amount of digital informa)on being produced is increasing . . .

n A total of 1.5 Exabytes of unique informa)on was produced in 1999. This
grew to 5 Exabytes by 2003.(1)
n IDC calculated that from 2006 to 2012 the amount of data generated
increased from 161 to 2837 Exabytes! (2)
• This informa)on is coming from a wide variety of sources . . .

Twi_er averages 6,000
Facebook’s 1.2 billion ac)ve daily tweets posted every second!
users post 300 million photos and
4.5 billion “likes” a day!
Google processes over 40,000

search queries every second
on average!
MIT Center for 1. Lyman, P. and Varian H., (2003) “How Much Informa)on”
Transportation & Logistics 2. Gantz, J. and Reinsel, D., (2012) “The Digital Universe in 2020” 2
To include supply chains . . . .
Amazon sells more than 480 million unique items to 244
million customers!
Walmart handles more than 1 million customer transac)ons every

hour, which is imported into databases es)mated to contain more
than 2.5 petabytes of data.
Every day, on average, UPS delivers over 20 million packages and

documents to more than 8.4 million delivery points using over
100,000 delivery vehicles.
Macy's adjusts pricing in near-real )me for 73 million

items, based on demand and inventory, every day.
MIT Center for

3
Other contribu)ng factors . . .
• Sensors – genera)ng ever more data
n Automobiles have 60-100 sensors today and will increase to over 200
n Smart packaging for shipment of perishable items captures temperature,
humidity, etc. while in transit
n Automated diagnos)c tests on computers and servers
• Internet of Things – machines communica)ng with machines
n There are ~6.4 billion connected things in use worldwide in 2016.
n Gartner forecasts that number will grow to ~21 billion by 2020.
n Es)mates of 170 million wearable devices online by 2019.
• Regula)ons – increasing the informa)on required to store
n 2013 Drug Supply Chain Security Act (DSCSA) requires “serialized numeric
iden)fier” (SNI)
w Example (Basta): Transac)on records for a single batch of 10,000 units
§ Pre-Serializa)on: 4 transac)ons (Create batch, Pick pallet, Ship pallet, Send ASN) ~ 2KB data
§ Post-Serializa)on: 60k transac)ons (10k each for Provisioning, Commissioning, Aggrega)ng, Picking,
Shipping, and Sending SNI data) ~ 11 MB data
MIT Center for Source: Basta, N. (2016) “Pharma Traceability: Year in Review”
4
It is not just the size, though . . .
Complexity is driven by Volume, Velocity, Variety, and Veracity!

MIT Center for
Transportation & Logistics Source: Rozados, I. and B. Tjahjono, (2014), “Big Data Analy)cs in Supply Chain Management” 5
Example: TinyCo
MIT Center for

6
TinyCo
• You have recently been hired by TinyCo within their Supply Chain
Management group.
• As your first assignment, you have been asked by your boss, Edith, VP Supply
Chain, to help understand the demand and forecas)ng processes. As part of
your learning, she has told you to look into a specific store, meet with the
store manager, and report back to her. Other members of your team will be
exploring other stores.
• You have asked the store managers for data on their sales for the last 3 years.
The person in charge of the data informs you that all of the stores are closed
on Sundays. He also has sent you a spreadsheet with the following email
message:
Welcome to ToyCo! I have a1ached the data that you asked for in the
spreadsheet. It has six tabs: sku masterand separate tabs for each of the
following stores: 312, 323, 415, 521, and 632. The store tabs contain all of
the daily sales informaHon for the last two years (that is all I could find) for
that respecHve store. It lists the database or transacHon ID, date, the SKU,
quanHty sold that day, and total revenue from those sales. The sku master tab
contains some informaHon on each SKU, such as its name, weight, cube, unit
cost, etc. This was really hard to get – so I hope that this is all you need.
MIT Center for

Transportation & Logistics Adapted from Blanco, E. (2006) SCM.250 Materials 7
TinyCo Data
SKU Master Table
Department Class Style Color SKU Cost Price Vendor
800 05 20 02 8000520021 7.50 9.99 MA Excellent Products
731 24 55 53 7312455530 14.50 22.99 GA General Wholesales
5001 201 12 4 50012011240 2.50 7.99 China Imports
5001 201 12 5 50012011250 7.50 9.99 China Imports
5001 201 13 4 50012011341 8.00 6.99 China Imports
5001 300 01 1 50013000110 6.50 12.99 China Imports
Transac)on File for Store 312

Unit Dollar
DB_ID SKU Store Date Sales Sales
79444 50012011250 312 8/3/14 3 29.97
79445 50012011250 312 8/4/14 2 19.98
79446
79447
50012011250
50012011250
312
312
8/5/14
8/6/14
5
3
49.95
29.97
Edith, ini)ally, wants
79448
79449
50012011250
50012011250
312
312
8/7/14
8/8/14
7
4
69.93
39.96
you to characterize the
79450 50012011250 312 8/10/14 4 39.96 total sales in both
79451 50012011250 312 8/11/14 4 39.96
79452 50012011250 312 8/12/14 1 9.99 units and dollar value.
79453 50012011250 312 8/13/14 3 29.97
79454 50012011250 312 8/14/14 4 39.96
79455 50012011250 312 8/15/14 1 9.99
79456 50012011250 312 8/17/14 2 19.98
... ... ... ... ... ...
MIT Center for
8
Recall How to Characterize a Distribu)on
• Central Tendency
n Mode – value that appears most frequently
n Median – value in the “middle” of a distribu)on, separa)ng the lower from
the higher half
n Mean (μ) – sum of values mul)plied by their probability (expected value
• Spread
n Range – maximum value minus minimum value
n Inner Quar)les – 75th percen)le value minus the 25th percen)le value
n Variance (σ2) - expecta)on of the squared devia)on around the mean
n Standard Devia)on (σ) - Square root of the variance
n Coefficient of Varia)on (CV) – Standard devia)on over the mean = σ/μ
n n 2 n 2
E[X] = x = µ = ∑ pi xi Var[X] = σ = ∑ pi ( xi − x ) = ∑ pi ( xi − µ )
2
i=1 i=1 i=1
MIT Center for

9
TinyCo Con)nued
MIT Center for

10
TinyCo – Store 312 Sales Summary
I am guessing many of you had
results something like this . . .
Unit Dollar
Minimum -6 -59.94 So . . . does anything look strange here?
25th Pctl 2 15.98
Mode 2 29.97 • How can I have nega)ve sales?
Median 3 31.99 • Why is my average so much larger than the median?
Average 108.08 1,097.78 • Why is the maximum so large?
75th Pctl 5 68.97 • Whoa! That is a huge coefficient of varia)on!
Maximum 60000 599400
Range 60006 599460
InnerQuar)le 3 52.99 Big warning flags and sirens
Variance 4,165,470 415,683,414 should be going off in your head!
Std. Dev. 2,041 20,388
CV 18.88 18.57
• All mathema)cal methods always assume clean data!

• As you describe data always keep in mind the following ques)ons:
• Is the data clean?
• Is the data complete?
• What assump)ons are you making about the data?
• Are the results making sense? How can I check?
MIT Center for
11
Cleaning Data is Never an Aterthought!!!
• Always plan enough )me for basic data checks
n Garbage in, garbage out
• Typical checks
n Invalid values - nega)ve, text, too small, too big, missing
n Mismatches between related data sets - # of rows, # of cols
n Duplica)on – unique iden)fiers
n Human error – wrong dates, invalid assump)ons
n Always explore the outliers – they are the most interes)ng!
• Be organized
n Versioning
n Keep track of data changes
MIT Center for

12
TinyCo – Store 312 Sales Summary – a 2nd Look
Looking a li_le closer . . .
Issue Ac;on
21 records with nega)ve sales Probably returns (Check on TinyCo Policy) keep in for now
34 records with zero value Removed
50 blank sales records (null) Removed
30 records with non-numeric data Removed
11 records with very large values Appear to be mul)plied by 10,000 – probably fat fingered –
flag and correct to right order of magnitude
3 duplicated entries Removed duplicates (2001601, 2001453, 2001331)
321 records w/ non-matching SKUs Flag and keep for now (8 @ 50*12011250 and 313
@50012011340) – Check for SKU changes
• Some Tips
• Always log issues and ac)ons – keep an audit mindset
• Never completely delete data – keep a complete data archive
• Create a “clean data set” to use in analysis
• Always validate decisions with process owners
• Use graphs (Sca_er Plots, Histograms, etc.)
• Use pivot tables (if you must stay in spreadsheets)
MIT Center for
13
TinyCo – Store 312 Sales Summary – a 2nd Look
Ater some ini)al cleaning, my summary is as follows . . .
Unit Dollar
Minimum -6 -59.94
25th Pctl 2 15.98
Mode 2 29.97
Median 3 31.99
Average 5.83 76.44
75th Pctl 5 68.97
Maximum 42 699.93
Range 48 759.87
InnerQuar)le 3 52.99
Variance 82 14,219
Std. Dev. 9 119
CV 1.55 1.56
• But this is barely the )p of the iceberg on what we want to do with this data!
• How do the different SKUs behave?
• How do sales differ by month? by week? by day of week?
• Are sales prices per SKU consistent?
• Are there trends in sales over )me?
• Do the returns (nega)ve sales) correlate to other sales?
• How are sales of the different SKUs related to each other?
MIT Center for
14
Querying the Data
MIT Center for

15
TinyCo II
• You are s)ll working for TinyCo. Edith would now like to
understand on how the the products sourced from MA Excellent
Products behave over )me.
• Specifically, she would like to know the following for those
selected SKUs:
n Which month has the highest average sales in dollars for store 312?
n Which specific week had the highest sales in dollars for store 312?
n Are the sales (in units) between the different SKUs from MA Excellent
Products correlated for store 312?
n How does the profit margin for sales in store 312 change over the )me
period?
n TinyCo is thinking of running a one day promo)on each week – which day
of week makes the most sense for store 312?
n How do these SKUs behave differently in the other stores?
How would you answer these queries in a spreadsheet?

MIT Center for
16
Querying the Data - Spreadsheets
• You can do this analysis in a spreadsheet, but it takes a
li_le bit of work.
• Pivot Tables can help
n Data summariza)on tool found in LibreOffice, Google Sheets,
and Excel
n They automa)cally sort, count, total or average the data
stored in one table or spreadsheet, displaying the results in a
second table showing the summarized data.
n Very useful in tabula)ng and cross-tabula)ng data
MIT Center for

17
Pivot Tables
Unit Dollar
DB_ID SKU Store Date Sales Sales DOW WOY Month Year Week Year Month Year
79568 50012011250 312 12/30/14 7 699.93 3 53 12 2014 201453 201412
85081 50013000110 312 5/26/16 42 545.58 5 22 5 2016 201622 201605
89834 8000451112 312 8/8/15 42 545.58 7 32 8 2015 201532 201508
89899 8000451112 312 11/14/15 42 545.58 7 46 11 2015 201546 201511
89946 8000451112 312 1/11/16 42 545.58 2 3 1 2016 201603 201601
89969 8000451112 312 1/30/16 42 Sum545.58
of 7 5 1 2016 201605 201601
89982 8000451112 312 2/13/16 42 Dollar
545.58 7 7
Sales Column Labels 2 2016 201607 201602
90010 8000451112 312 3/12/16 42 Row 545.58
Labels 7 11
7312455520 3 8000451112
2016 8000520021
201611 201603
Grand Total
90028 8000451112 312 4/1/16 42 545.58
1 6 14
$1,477 4 2016 201614 201604
$17,387 $2,018 $20,882
90071 8000451112 312 5/24/16 42 545.58 3 22 5 2016 201622 201605
90093 8000451112 312 6/21/16 42 2
545.58 3 $1,683
26 6 $12,573
2016 $2,308
201626 $16,564
201606
90105 8000451112 312 7/1/16 42 3
545.58 6 $2,239
27 7 $14,211
2016 $2,208 201607
201627 $18,658
90115 8000451112 312 7/9/16 42 545.58
4 7 28
$1,693 7 2016 201628 201607
$15,387 $2,458 $19,537
90158 8000451112 312 8/20/16 42 545.58 7 34 8 2016 201634 201608
5 $1,801 $13,401 $2,887 $18,090
6 $1,553 $11,397 $2,757 $15,708
7 $2,058 $16,767 $2,677 $21,503
8 $1,487 $15,882 $2,537 $19,907
9 $1,551 $7,800 $1,499 $10,850
10 $2,327 $7,797 $1,838 $11,962
11 $1,204 $9,432 $2,138 $12,773
12 $1,697 $10,285 $2,068 $14,050
Grand
Total $20,774 $152,318 $27,393 $200,484
MIT Center for
18
The Big Data Challenge
MIT Center for

19
The Opportunity . . .
• The Digital Disrup)on is here . . .
n World’s largest taxi company owns no taxis (Uber)
n Largest accommoda)on provider owns no real estate (Airbnb)
n Largest phone company owns no telco infrastructure (Skype, WeChat)
n World’s most valuable retailer has no inventory (Alibaba)
n Most popular media owner creates no content (Facebook)
n Fastest growing banks have no actual money (SocietyOne)
n World’s largest movie house owns no cinemas (Nezlix)
n Largest sotware vendors don’t write apps (Apple/Google)
• Data Analy)cs have become pervasive . .

n Retail and CPG - Demand and sales forecas)ng, Promo)ons, On-line pricing and product
recommenda)ons
n Financial Services – Underwri)ng, Credit risk, Fraud analysis, Li)ga)on mi)ga)on
n Telecommunica)ons - TV networks, Product subscrip)ons
n Marke)ng - Direct marke)ng, email, mobile ads, SEO, CRM, Loyalty programs, Customer
acquisi)on and reten)on,
n Distribu)on network and supply chain – Real-)me track & trace, Micro-segmenta)on,
Predic)ve cost es)ma)on, Asset monitoring (digital twins), Serializa)on
MIT Center for Source: [www.ibmforentrepreneurs.com] and Cassa, C. (2016) Course notes SCM.264
20
The Challenge . . .
“It’s an absolute myth that you can send an algorithm
over raw data and have insights pop up.”
Jeffrey Heer, co-founder of Trifacta
“Facts are simple and facts are straight

Facts are lazy and facts are late
Facts all come with points of view
Facts don't do what I want them to
Facts just twist the truth around
Facts are living turned inside out”
The Talking Heads
There is no magical data science machine turning data into insights!

• Data is messy – always requires lots of cleaning and usually some programming
• Data is usually siloed – need to combine data from mul)ple disparate sources
• Data is big and ge~ng bigger – standard techniques and tools (i.e., spreadsheets)
are no longer sufficient
MIT Center for Source: Cassa, C. (2016) Course notes SCM.264

21

SC0x 2T2023 Module 1 Annotated

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

SC0x 2T2023 Module 1 Annotated

Uploaded by

Copyright:

Available Formats

CTL.

SC1x -Supply- Supply

Introduction to Logistics & Supply

MIT Center for

1. What is a supply chain?

• The primary purpose of a SC is to satisfy customer needs.

• Supply Chains try to maximize the total value generated

“Supply Chain Management deals with the management of materials,

“Call it distribution or logistics or supply chain management. By

n Logistics management is that part of supply chain management that

n Supply chain management encompasses the planning and

Source: Used by permission, Council of Supply Chain Management Professionals, http://cscmp.org/.

SRM ISCM CRM

Supplier Relationship Internal Supply Chain Customer Relationship

Different macro processes should be aligned

Plan P1 Plan Supply Chain

P2 Plan Source P3 Plan Make P4 Plan Deliver P5 Plan Returns

Source Make Deliver

S2 Source MTO Products D2 Deliver MTO Products

S3 Source ETO Products M3 Engineer-to-Order D3 Deliver ETO Products

Objective: Deliver at lowest transport cost

Objective: Deliver at lowest transport cost

• Politics and Power of Players - Who wins?

• Visibility – Who can see what and how quickly?

• Increased complexity – Why is it getting harder?

• Global operations – Why don’t we ever close?

Evolved into being both a Bridge and a Shock Absorber

Data Source: U.S. Energy Information Administration, http://www.eia.gov/petroleum/gasdiesel/diesel_proc-methods.cfm

US Inventory Holding Cost US Transportation Cost Total US Logistics Cost

Key Skills Technically Narrow Coordination

Influence Local & Controllable Across entire Supply Chain

Leadership Hierarchical, Direct, Influential, Indirect, ‘Soft’

Measurement Single Focus Multi-faceted

Technology Isolated Optimization Visibility & Coordination

Supply Chain Management is a growing and evolving discipline.

Questions, Comments, Suggestions?

MIT Center for

n How many facilities should I open and where?

Human is part of Human is external to

Real Operational Analytical

Source: Bradley, Hax, Magnanti 1977 3

Deterministic Simulation Linear Programming

Source: Bradley, Hax, Magnanti 1977 4

Model Functional Independent OR/MS

Predictive unknown, known or under Regression Analysis,

Prescriptive known, known or under Classic Opt., LP, MILP,

Parabola - Polynomial function of degree 2

• When a>0, the function is convex (or concave up)

n Values of x for when y=0 x

4 ± (−4)2 − 4(3)(2) 4 ± 16 − 24 4 ± −8 (i) y = 2x2

w cost =f(# made) = 500,000 + 75x b

n Demand for iWidgets is also a linear x

function of the price, p: y

w unit sales = f(price) = 20,000 – 80p

1.00 if 0<b<1, then function

y(b=1) y(b=1.5) y(b=.5) y(b=-1)

• Exponential functions have very fast growth

100 = 10x log10(100) = x x=2

Properties of Logarithms Examples:

Adapted from http://www.purplemath.com/modules/logs.htm 24

n F=2P=P(1+r)n which reduces to 2=(1+r)n=(1.07)n

n We can transform this by taking the ln or log of both sides:

Number of Years to Double in