Applied Analytics for Operations,

Supply Chain and Logistics

Forecasting for Inventory Control

Devon Barrow
Email: [email protected]
Group assignment
Group assignment: Highlights

 The group assignment is now available.

 Deadline: 22nd February 2024, 12.00 p.m.

 It is a case study analysis: you choose the case

 The analysis should be submitted in the form of a group presentation

 The submission format is an audio recorded PowerPoint Presentation

to be uploaded via Canvas. Use the group presentation template.
Group assignment: Highlights

 Choose an operation and perform an analysis of its inventory

management approach e.g. a bakery or pharmacy or warehouse or
supermarket

 Choose an operation to which you can obtain access to data and

whose inventory management approach will be of interest to all
members of the group.
Group assignment: Highlights

It is recognised that data for this analysis may be challenging to obtain. As such you
should exhaust all effort to obtain data as follows:

1. Secondary data e.g. business press, white papers, company reports, etc.
2. Primary data collection through own observation or interviews with key
personnel from the operation. Do not engage in any survey data collection as
this would require ethical approval.
3. Simulated data based on sensible assumptions e.g. demand for a product/item
may be assumed based on the size and location of the operation. Price for a
product of item may be assumed based on a similar product.
4. A sensible combination of 1), 2) and 3)
Group assignment: Highlights

 We will run a group peer assessment. This will be communicated

shortly.
 All Students will evaluate their group members (except themselves)
using an excel spreadsheet form.
 You will complete the form individually using Excel and submit a copy
to a designated page.
 More information will be provided in Week 3 via Canvas.
Unit 3 learning outcomes

 Generate forecasts in the form required for use in inventory control models.
 Evaluate inventory models of relevance to low demand uncertainty e.g. EOQ
which assumes constant demand.
 Evaluate inventory models of relevance to high demand uncertainty e.g. the
established benchmark Order-up to Inventory Policy.
 Evaluate the performance of different forecasting methods in terms of both
forecasting accuracy and inventory performance
Office hours

Dr Devon Barrow, Room 1124

Monday 11.30-1.30pm: Face to face meetings, Room 1124
Wednesday 1.30-3.30pm: Online via Zoom, Email to schedule
Elements of Inventory Management
 A system to keep track of inventory
 A reliable forecast of demand
 Knowledge of lead times: time interval between ordering and receiving the order
 Reasonable estimates of:
– Holding costs – cost to carry an item in inventory for a length of time
– Ordering costs – costs of ordering and receiving inventory
– Shortage costs – costs when demand exceeds supply
 A classification system
Inventory Classification System
 Classifying inventory according to some measure of importance and allocating control
efforts accordingly.

 A – very important
 B – moderately important
 C – least important
ABC System
 Different inventory management decision rules are needed for different classes of inventory.
– Some stocked items are more important than others
 A high usage rate

 If they ran out, customers would be disappointed.

– Others might be of particularly high value

 Excessively high inventory levels would be particularly expensive.

 One common way of discriminating between different stock items is to rank them by their usage
value (usage rate x unit value)
ABC System
 Not generally economical to give same degree of attention to all items in a multi-product system.

 Most real inventory systems are found to fit these approximate proportions
– Class A: 20% items, 80% value
– Class B: 40% items, 15% value
– Class C: 40% items, 5% value

 This classification of items into groups is known as ABC analysis or Pareto analysis

 Highest priority (and degree of control) to Class A items, and less priority to Class B and C – more economic
use of resources.
Inventory Control Systems
 Period System
Physical count of items made at periodic
intervals

 Continuous Inventory System

System that keeps track of removals from
inventory continuously, thus monitoring current
levels of each item
Inventory Control (Policies)

 Continuous Review and Reorder Point

 Periodic Review and Order Up-to Level
 Periodic Review, Reorder Point and Fixed Order Quantity
Key terms
On-hand inventory

•Inventory physically available for a client to buy.

Backorders

•Backlog of open orders that are not yet fulfilled. This happens when you do not have enough on-hand inventory
to fulfil orders directly and the orders are not lost.

Net inventory

•Inventory level including available on-hand inventory and in-transit inventory, minus backorders, orders not yet
shipped, etc.

In-transit inventory

•Goods ordered from a supplier but not yet available in our warehouse for our clients to buy. These goods are
considered to be in-transit between two warehouses (or in pipeline).
Continuous Review and Reorder Point

 Order products as soon as the net inventory reaches a threshold

(or goes below).
 We order a predetermined number of units from our supplier.
 This threshold is called the reorder point or ROP.

Notation

• The fixed reorder point policy is noted (s, Q), with s the reorder point and Q the
fixed order quantity.
Continuous Review and Reorder Point

• When the stock level reaches 3 pieces,

I order 10.

• Here the fixed reorder point is 3, and

the order quantity is 10

• Even if you only have 2 pieces in

stock, you will still only order 10
pieces.

• In the fixed reorder point, the order

quantity (10 in this case) is always
fixed.
Continuous Review and Reorder Point:
Examples
 Online Retailers: An online retailer selling electronic gadgets might use a fixed reorder point policy for a high-
demand product. When the inventory drops below the set point, an automated system triggers the immediate
replenishment order to restock the item.

 Automotive Parts Store: In an automotive parts store, a fixed reorder point policy could be applied to critical
components like brake pads. When the inventory level falls below the reorder point, the store initiates the
replenishment process to ensure that these essential items are always in stock.

 Manufacturing Plants: Manufacturing facilities often implement fixed reorder point policies for raw materials. When
the stock of a particular raw material reaches the reorder point, the procurement team is notified to place an order to
maintain uninterrupted production.

 Pharmacies: Pharmacies may use a fixed reorder point policy for prescription medications. When the quantity of a
specific medication drops below the predetermined reorder point, the pharmacy places an order to restock the
medication and meet customer demand.
Continuous Review and Reorder Point:
Advantages

 This policy is safe (i. e., the risk of being out-of-stock is low) as it assumes you
can make an order whenever you need to. It is therefore a good policy for
expensive and/or important items that need to be monitored closely.
 You can optimize the order quantity based on some (often obvious) constraints
or costs. For example, you might get a rebate if you order a full pallet or a
truckload. With such a policy you are sure to get the reduction each time you
make an order.
Continuous Review and Reorder Point:
Limitations

 It won’t allow you to group into a single order different items with a single
supplier. The policy assumes that each client can only buy one product at a
time, so that the reorder point will always be perfectly reached.
 It assumes that a client can make an order with its supplier at any time.
Periodic Review and Order Up-to-Level

 Order products periodically, based on a fixed schedule and on an up-to level.

 At the beginning of any review period, we will order enough products to bring our
net inventory to the up-to level.
 The elapsed time between two consecutive orders will always be the same (but
the order quantity will always change)..

Notation

• The fixed review period policy is often noted (R, S), with R being the fixed
review period and S the up-to level.
Periodic Review and Order Up-to-Level

• Fixed review period policy (with

immediate replenishment).

• Order up to 13 every 5 days

• Fixed review period R = 5 days

• Order-up-to S = 13
Periodic Review and Order Up-to-Level:
Examples
 Bookstore: Imagine a bookstore that conducts a periodic review of its inventory every
month. During this review, the store checks the stock levels of all book categories. If any
category's inventory falls below a predetermined threshold, the store places an order to
bring it back up to the desired level.

 Electronics Store: An electronics store may perform a monthly inventory review. If the
stock of a popular smartphone model is lower than the predetermined level, the store places
an order to replenish the inventory and meet customer demand.

 Clothing Boutique: A boutique may conduct a periodic review of its clothing inventory
every two weeks. If the inventory of a particular clothing line is below the specified level, the
boutique places an order to ensure a diverse selection for customers.
Periodic Review and Order Up-to-Level:
Advantages

 The periodic replenishment policy (with an order up-to level) is the most
common inventory policy because it allows businesses to group their
orders with each of their suppliers.
 Often forced onto supply chains by the use of material replenishment
planning. These tend to follow a predefined schedule, often daily or
weekly, resulting in the implicit use of a periodic review policy.
Periodic Review and Order Up-to-Level:
Limitations

 This policy is riskier, due to the blind spot it creates. You cannot order
in-between two review periods.
 The order quantity will vary at each order potentially causing disruption
to smooth operational flow.
Periodic Review, Reorder Point and Fixed Order
Quantity

 This policy will consist of making orders of fixed quantity Q, based on a fixed
schedule, if the inventory level reaches a threshold s.
 Offers the convenience of a fixed order quantity, and a fixed review period.

Notation

• The policy is denoted (R, s, Q), where R denotes the fixed review period, s the
fixed reorder point, and Q the fixed order quantity.
Periodic Review, Reorder Point and Fixed
Order Quantity

• Period review with reorder point

and fixed order quantity

• Order 10 pieces every 4 days

only if inventory level reaches 5
pieces.

• Fixed review period R =4

• Fixed order quantity Q=10

• Reorder point s =5
Periodic Review, Reorder Point and Fixed Order
Quantity: Example
 Office Supplies Replenishment for a Small Business: Imagine a small business that regularly
replenishes office supplies using a combination of Periodic Review, Reorder Point, and Fixed Order
Quantity strategies.

 Periodic Review: The store decides to conduct inventory reviews every two weeks. This means that at the
end of each two-week period, they will assess the current inventory levels.

 Reorder Point: Based on historical sales data and lead time analysis, the store determines that their
Reorder Point (ROP) is when the inventory level drops to 50 units. This is the minimum level at which they
need to reorder to avoid stockouts during the lead time.

 Fixed Order Quantity (FOQ): The store has decided to order a fixed quantity of 100 units each time they
place an order. This ensures that they take advantage of bulk purchasing discounts and minimize order
processing costs.
Periodic Review, Reorder Point and Fixed Order
Quantity: Example
 Day 1: The store receives a shipment of 100 units of a particular product. The
inventory level is now 100 units.
 Day 10: The sales have been steady, and the inventory level is now 60 units.
 Day 14 (End of the 2-week Period - Periodic Review): The store conducts a review
and sees that the inventory level is below the Reorder Point (ROP) of 50 units. It
triggers the need to place an order.
 Day 15: The store places an order for a Fixed Order Quantity (FOQ) of 100 units.
 Lead Time (Delivery Time): The lead time for the supplier to deliver the order is 5
days.
Periodic Review, Reorder Point and Fixed Order
Quantity: Example
 Day 20: The inventory is now at 30 units (60 units remaining from the previous
inventory, minus 30 units sold in the last 5 days).
 Day 21: The order of 100 units arrives from the supplier, bringing the inventory back
to 130 units (30 units + 100 units from the new order).

 This cycle repeats, and the store continues to review, reorder, and receive shipments
based on the Periodic Review, Reorder Point, and Fixed Order Quantity methods.
 This approach helps the grocery store maintain an optimal inventory level, reduce the
risk of stockouts, and take advantage of cost efficiencies associated with ordering in
fixed quantities.
Periodic Review, Reorder Point and Fixed Order
Quantity: Advantages

 The order quantity is always constant, allowing transportation and

packaging to be optimized (e. g., full pallet, full truckload).
 The order is always made at a predefined time slot (e. g., at the end of
the day/week), allowing smooth operations and the ability to group
orders with a supplier.
Periodic Review, Reorder Point and Fixed Order
Quantity: Limitations

 The risk is high with this kind of policy. The review period blocks an
order being made when it may be needed.
 An order is only made if stock reaches a certain threshold.
 This extra risk will need to be compensated for by an extra amount of
safety stock, resulting in higher inventory levels and costs.
Inventory Decisions
Inventory Decisions

 How much inventory do we need?

How much should be ordered?
When should the order be placed?

 The role of forecasting

When demand is constant and leadtime is constant
When demand is uncertain and leadtime is constant
Inventory Decisions

Transaction
cost k Backorder
Order cost bt
quantity Q
Demand
Suppliers Clients
Instantaneous

Warehouse
Holding cost h

Supply chain Costs

How much should we order?

 Aim: The right amount of stock

Too much
Cost of storing stock: holding costs, risk of obsolescence
Capital cost: loss of interest of capital tied up
Reduced working capital
Too little
Stock outs, lost sales, disappointed customers
Higher costs: emergency orders, increased order frequency
 Goal: Balance cost of too much versus too little
Economic order quantity
How much should we order?
Slope = demand rate (steady
and predictable)
Inventory

Q Ave. Inventory
Q/2

Q/D Time

Frequency of orders = D/Q

How much should we order? No
Forecasting
Required
Slope = demand rate (steady
and predictable)
Inventory

Q Ave. Inventory
Q/2

Q/D Time

Frequency of orders = D/Q

 When should the order be placed? No
Forecasting
Constant Demand Required

Inventory

ROL

Lead time Time

ROP Notation

• The fixed reorder point policy is noted (s, Q), with

s the reorder point and Q the fixed order quantity.
 Holding costs = holding costs/unit x average inventory
= Ch x Q/2

The Basic EOQ Model Ordering costs = ordering cost x number of orders per
period
= Co x D/Q

Costs Total cost, Ct = ChQ/2 + CoD/Q

What is the optimal order quantity Q?

The quantity which minimizes total cost.

Total Costs

Holding costs

Order costs

EOQ Order Quantity

EOQ Example
 A large bakery buys flour in 25-pound bags. The bakery uses an average of 4,860
bags a year. Preparing an order and receiving a shipment of flour involves a cost of
£10 per order. Annual holding costs are £75 per bag.

1. Determine the EOQ

2. What is the average number of bags on hand?
3. How many order per year will there be?
4. Compute the total cost of ordering and carrying flour.
5. If ordering costs were to increase by £1 per order, how much would that affect the
minimum total annual cost?
EOQ Example Solution
 D= 4,860 bags/ yr
 Co = £10
 Ch = £75

1. Q = EOQ = SQRT(2*DC*Co / Ch) = SQRT(2*4,860*10) / 75) = 36 bags

2. EOQ / 2 = 36/2 = 18 bags

EOQ Example Solution
 D= 4,860 bags/ yr
 Co = £10
 Ch = £75

3. D/EOQ = 4,860 bags /36 bags per order = 135 orders

4. TC = EOQ/2 * Ch + D/Q * Co = 36/2*75 + 4,860/36*10 = £1,350 +

£1,350 = £2,700
EOQ Example Solution
 D= 4,860 bags/ yr
 Co = £10
 Ch = £75

5. Using Co = 11, EOQ = SQRT(2*DC*Co / Ch) = SQRT(2*4,860*11) / 75)

= 37.757 bags
TC = 37.75/2*75 + 4,860/37.75*10 = £1,415.89 +
£1,415.89 = £2,831.79  increase of £131.79
Some criticism of EOQ

 Assumption of steady demand

 Nature of costs: importance of order versus holding costs (trade-off)
 Assumes that costs cannot be affected: Instead of asking “What is the optimum order
quantity?” ask how can I change my operations to reduce inventory levels?
 Should the cost of inventory be minimized?
The Role of Forecasting
 When should the order be placed?
Uncertain Demand
Inventory
Normal distribution
of lead-time usage

ROP
ROL

Time

Lead time
 When should the order be placed?
Uncertain Demand
Inventory
Normal distribution of
lead-time usage.
Max. probably
demand during
leadtime.

ROP ROP ROP

ROL

Time
Lead time Lead time Lead time
 When should the order be placed? Forecasting
Required

Uncertain Demand
Inventory If demand was
constant,
Normal distribution of
expected demand
lead-time usage.
during leadtime.
Max. probably
demand during
leadtime.

Q
ROP ROP ROP
ROL
0

Time
Lead time Lead time Lead time
 When should the order be placed? Forecasting
Required

Uncertain Demand
Inventory If demand was
constant,
Normal distribution of
expected demand
lead-time usage.
during leadtime.
Max. probably
demand during
leadtime.

Q
ROP ROP ROP
ROL
0 Safety Stock Safety Stock Safety Stock

Time
Lead time Lead time Lead time

Safety Stock: reduces risk of stockout during leadtime.

The role of forecasting
Safety stock: stock that is held in excess of expected demand due to variable demand rate and/or lead time.

The ROP based on a normal distribution

of lead time demand

Risk of a stockout

Expected
Demand

Quantity
Safety ROP

Stock
 Replace average

The role of forecasting demand over leadtime

with sum of forecast
demand over
leadtime.

𝑑̅ × 𝐿𝑇
…… (1)
 𝑑̅ = average demand per day (or week or month or year)
 LT = expected lead time

 We will use instead the sum of all the forecasts over the lead time rather than (1) above. We
the forecast horizon to lead time.

𝐹
…… (2)

 𝐹 = Forecast of demand per day (or week or month)

The role of forecasting

𝜎 =𝜎 𝐿𝑇

∑ (𝑑 −𝑑̅ ) ∑ (𝑑 )
𝜎 = = − 𝑑̅
𝑛 𝑛

 𝑑̅ = average demand per day (or week or month or year)

 LT = expected lead time
 Replace average

The role of forecasting demand with

forecast to get
standard deviation
of forecast
demand.

𝜎 =𝜎 𝐿𝑇

∑ (𝐴 −𝐹 )
𝜎 =
𝑛

 𝐹 = Forecast of demand per day (or week or month or year)

 LT = expected lead time
The role of forecasting
Safety stock: stock that is held in excess of expected demand due to variable demand rate and/or lead time.

The ROP based on a normal distribution

of lead time demand

Risk of a stockout

Expected
Demand

Quantity
Safety ROP

Stock

∑ (𝐴 −𝐹 )
𝐹 + 𝑛
× 𝐿𝑇
The role of forecasting:
Re-order Point Quantity Calculation
𝜎 = 𝑑̅ × 𝐿𝑇 + 𝑧𝜎 𝐿𝑇

 Where z corresponds to the standard normal distribution and the probability of not having a
stock out). For example, z = 1.645 corresponds to a probability of 95% of no stockout.

∑ (𝐴 −𝐹 )
𝐹 + × 𝐿𝑇 × 𝑧
𝑛

Cumulative Demand Standard deviation of Forecast Square root of lead Safety Factor
(Forecast) during lead time errors (=In-sample RMSE) time (z=1.645 provides
(=Forecast horizon h) 95% probability of no
stock out)
The Normal Distribution
Normal distribution

The entire family of normal probability distributions is

defined by its mean and its standard deviation .

Standard Deviation

x
Mean
Normal distribution

68.26% of values of a normal random variable

are within +/- 1 standard deviation of its mean.

95.44% of values of a normal random variable

are within +/- 2 standard deviations of its mean.

99.72% of values of a normal random variable

are within +/- 3 standard deviations of its mean.
Normal distribution
99.72%
95.44%
68.26%

x
m
m – 3s m – 1s m + 1s m + 3s
m – 2s m + 2s
The Standard Normal distribution
99.72%
95.44%
68.26%

Z
m
– 3s – 1s 1s 3s
– 2s 2s
The Standard Normal distribution
For stock control we are not interested in both sides. So only the right side:

95.44%
95.44%

Z
0 3s
– 3s – 1s 1s
– 2s 1.689s 2s
The Standard Normal distribution
A more common no stock out percentage is 95%, leaving a 5% chance of a
stockout. The Z score value is 1.645.

Area = 0.9500

Area = 0.0500

z
0 z.05=1.645
The Standard Normal distribution
Here are the z-values for some common safety stock factors (probabilities):

(Desired) Risk of Stockout Z-Value

No Stockout Risk (Service Level)
75% 25% 0.674

90% 10% 1.282

95% 5% 1.645

99% 1% 2.326
Next steps

Unit 4
Quiz: Fundamentals of Probability
Discussion Forum: Decision-making
 Please make a post
Workshop
 ABC (Pareto) Analysis
 Simulating an (R, S) inventory policy

Week 5 and 6 will be fully uploaded over the weekend.

Questions?