2021 Methods of Forecasting
2021 Methods of Forecasting
2021 Methods of Forecasting
Typically, a variety of forecasting methods are applicable to any particular type of supply chain scenario.
Smart supply chain planners use multiple methods tuned to perform well at different phases of the product
life cycle, chosen to best exploit the available historical data and degree of market knowledge. The key is to
pick the most effective and flexible methods and models, blend their best features, and shift between them
as needed to keep forecast accuracy at its peak.
In this paper we take a brief look at the three categories of forecasting models and the eight methods
that have produced superior results for Logility’s many clients in a variety of industries and market
conditions around the world. We also discuss how Multi-Variate Demand Signal Management can help you
incorporate internal and external demand data to improve forecast quality and uncover insights to make
better and faster decisions.
1
The Hierarchy of Supply Chain Metrics: Diagnosing Your Supply Chain Health
Refreshed: 15 November 2016 | Published: 26 June 2015
Analyst[s]: Debra Hofman
Table of Contents
Three Categories of Forecasting Models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Eight Forecasting Methods that Improve Supply Chain Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Best-fit Statistical Modeling. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Derived Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Modeling for Intermittent Demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Attribute-based Modeling. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Demand Sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Multi-Variate Demand Signal Management. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1 Qualitative models are experience-driven, relying on subjective inputs from knowledgeable personnel,
such as salespeople, account managers, and the like. This approach typically sets up formal procedures
for data review, and requires a consensus to determine the value of various forms of information.
Consensus among forecasters may be obtained by aggregating individual estimates or through
structured polling methods.
Derived models create new forecasts based on existing forecasts. When a new item’s forecast is thought
to be fundamentally the same as an existing item, characteristics can be used in creating the new
forecast, which may be factored up or down by a percentage. This preserves the overall trend and
seasonal characteristics of the item, providing a good starting point for the new item.
3 Hybrid models typically draw on historical demand information as a starting point, then use empirical
data to further refine the forecast.
Attribute-based models employ user-defined attributes to model new product introductions, seasonal or
fashion driven products, and product end-of-life retirement based on a demand profile.
Causal models use a causal relationship between a particular time series variable and other time series
factors to calculate the forecast. Causal techniques are useful in forecasting ‘lift’ during promotional
campaigns, where demand caused by promotional factors has an established relationship to base
demand. Additional factors can be used, such as end-cap displays, seasonality of the product, etc.
Factors are not additive but are used together to calculate the expected lift for your product.
In Logility’s experience working with more than 1,200 organizations ranging across dozens of industries,
eight specific forecasting methods stand out. Their unique strengths combine to deliver powerful, flexible
and accurate results.
1 Modified Holt is a best-fit statistical technique used when demand is trended, but does not vary by
[1]
the time of the year. A Holt-Winters variant is often used when demand is seasonal.
2 Moving Average is used for products whose demand histories have random variations, including no
[2]
seasonality or trend, or a fairly flat demand.
4 Modified Parent-Child is a derived model technique used to forecast products as a percent of the
[4]
forecast for another product [dependent demand].
5 Modified Croston is an intermittent demand technique used for products such as slow-moving parts
[5]
that have low demand or some zero demand periods.
6 The Demand Profile technique is attribute based. It employs user-defined attributes to model new
[6]
product introductions and product end-of-life retirement.
[8]
8 Demand Sensing techniques provide real-time visibility and insights into short-term demand,
and are enabled through down-stream data sources such as POS and syndicated scanner data and
by advanced technologies such as machine learning driven pattern recognition and natural
language processing algorithms, simulation, and optimization.
A modified Holt-Winters decomposition model with best-fit analysis can generate forecasts based on
demand history that incorporate trends and seasonal information. The method “senses” the amount of
history available for each time series or segment to create a basic model that best fits the history. Then it
uses the best combination of smoothing factors to enable the model to react to changing conditions going
forward without overreacting to anomalies in demand [such as unplanned seasonal events, transportation
disruptions, and so on].
For factors relating to seasonality, planners need the ability to weight the historical demand. Under the
assumption that the previous year is the best indicator of what will happen next year, most forecast
systems apply a higher weighting factor to the previous year’s demand, less to the year before and even
less to the years before that. But if the previous year was unusual in any significant way, the planner
must have the capability to change the historical weighting factors [so that the history two years ago
has more impact on the current forecast than last year, for instance] so as not to under- or over-forecast
the business.
Seasonal methods can be effective with less than 24 months of history; the minimum required is twelve
months. An effective approach for expected seasonal items with less than twelve months of history is to
assign a seasonal curve that has been captured from a similar item or item group.
A powerful best-fit statistical method should include flexible features such as trend, seasonal-with-trend,
moving average and low-level pattern fitting, as well as trend models for products with sporadic, low-
volume demand. The method should provide limiting and damping, as well as seasonal smoothing,
demand filtering, reasonability tests, tracking signals and tests for erratic nature that evaluate the validity of
each element, determining which are anomalous and should be filtered. These parameters give the planner
the flexibility to tune the process to best fit conditions at any element of the organization.
“Best fit” refers to the ability to change forecast methods as a product evolves.
The process may start out as a demand profile method, evolve to a modified
Holt-Winters method as the product becomes stable, and ultimately transition to a
demand profile method again as the product life cycle comes to an end.
Limiting Confidence Limits describe the spread of the distribution above and below the point forecast.
Removes random variation [noise] from the historical demand, enabling better identification
Smoothing of demand patterns [primarily trend and seasonality patterns] and demand levels. Results in a
closer estimate of future demand.
Applies various “weights” to each period to achieve the desired results. These weights are
Damping
expressed as percentages, and the total of all weights for all periods must add up to 100%.
Forecast error, viewed as the difference between forecast value and actual value, is usually
normally distributed. A Demand Filter is usually set to ±4 Mean Absolute Deviation against
Filtering
the forecast value. Whenever the deviation is more than that, the adequateness of the
forecast model should be reviewed by analyzing the actual data.
The difference between actual demand and forecast demand. Error can occur in two ways:
bias and random variation. Bias is a systematic error that occurs when cumulative actual
demand is consistently above or below the cumulative forecast demand.
Forecast Error Type 1 Bias is subjective and occurs due to human intervention.
Type 2 Bias is a manifestation of a business process that is specific to the product [for
instance, persistent demand trend and forecast adjustments don’t correct fast enough for
items specific to a few customers].
An important type of reasonability measure is the tracking signal, which can be used to
Reasonability monitor the quality of the forecast. There are several procedures used, but one of the simpler
Tests is based on a comparison of the cumulative sum of the forecast errors to the mean absolute
deviation.
“Best fit” refers to the ability to change forecast methods as a product evolves. The process may start out as
a demand profile method, evolve to a modified Holt-Winters method as the product becomes stable, and
ultimately transition to a demand profile method again as the product life cycle comes to an end.
The patterned variation looks at available history and classifies each demand element relative to those
around it. It classifies the periods into peaks, valleys, plains, plateaus, up-slopes and down-slopes. It
measures the duration of plateaus and plains, as well as the severity of peaks and valleys. It then conducts
pattern-fitting analysis to find regularity over time, attempting to fit the pattern to the history and
averaging for low and high points. The patterned forecast is put in context of future periods with the
average trend, and the pattern is re-evaluated using demand history of subsequent periods.
If no pattern is present, the unpatterned variation method attempts to use averaged highs and lows to
create a step-change forecast for future demand.
Both techniques permit zero demand to reside in the history, and will acknowledge such in the future
demand forecast. In forecasting for spare parts, for example, the demand is frequently low-level and spotty,
containing many periods of zero demand interspersed with low-level demand. This forecasting technique
allows patterns of zero demand to be forecast into the future.
Demand sensing involves the import of short-term demand data such as Point of Sale [POS] scan-based
data, point of use device data, weather data, or social media data on an hourly/daily basis to immediately
sense demand signal changes. Through advanced algorithms the statistical significance of demand changes
are evaluated and short term forecast adjustments made to drive short term supply chain responses.
The typical performance of demand sensing can reduce near-term forecast error by 30% or more
compared to traditional time-series forecasting techniques. However, since demand sensing usually
takes place inside of the supply planning time fence, the greatest value gained from demand sensing
is the ability to optimize inventory and resource deployment. Demand sensing can lead to a 5%—10%
improvement in customer service.
An evolving demand sensing trend is the use of Artificial Intelligence [AI] to automate the process of
analyzing Big Data to recognize complex patterns and to separate actionable demand signals. Used
together, machine learning and natural language processing algorithms can be used to analyze information
like social media text to determine the “Sentiment” of the text and to predict the impact of that sentiment
on demand. Today’s natural language processing algorithms have the ability to correctly categorize the
sentiment of most social media content. Machine learning algorithms can quickly learn the differences
between humor, sarcasm, irony and so on to improve categorization capabilities.
Companies are using social media insights today to make significant operational impacts.
● Evaluate the Health of a Brand—An understanding of how your target market feels about your
company, product and services through analysis of overall sentiment can provide valuable insight into
the health of your brand.
● Address a Crisis—Analysis of social sentiment might reveal a spike in negative posts and provide an
early warning to a potential product or service issue. Through alerts and analysis, the root cause of the
issue can be uncovered and corrected.
● Research the Competition—Social sentiment analysis can help you understand how to position against
the competition.
● Improve Demand Prediction—Companies can now use the ‘Voice of the Consumer’ to drive
improvements in forecasting and inventory positioning.
LOGILITY ADAPLINK
CRM ERP CDM Suppliers Customers
● Demand planners are better prepared to quantify variability in demand through increased visibility
downstream by capturing and analyzing external demand signals
● Planners can improve forecast accuracy by incorporating information from external demand signals and
predicting mid-term demand patterns
● Operations improve lead time because of the improved use of real-time forecast to drive enhanced
scheduling and procurement
● Sales can boost sales by predicting and reducing stock outs with demand pattern recognition
● Sales can Improve new product introduction through better customer insight at POS
● Operations can lower expediting and inventory costs
● Logistics build transport and deployment scheduling efficiency because of improved visibility of
customer requirements
Demand Forecasting
Multi-Variate Demand Sensing Management
Foundation of Demand Planning
Aggregate Time Series Demand
DemandSensing
Sensing
Augmentation
Orders/Shipments Integrates all relevant demand
Mid/Long-range Horizon signals to single source of truth Uses POS data
“What am I going to buy/make” Machine Learning with minimal latency to
know what is being
Market and external demand sold, where it is being
signals sold, and to whom
“How do I react to external Daily Forecasting
factors driving demand for
buy decisions” “What am I going to
ship”
Attribute-based methods that use demand profiles are often suited to new product introduction and end
of product life cycles, at times when reliable historical demand data is lacking or the available data is less
relevant.
At the more mature stages of the product life cycle, five different time-series statistical models come into
play, including modified Holt, Holt-Winters, moving average, and intermittent or low demand, whether
patterned or unpatterned. These models are used to create retrospective forecasts that cover prior periods
[typically three years] of documented demand.
To prevail in a business economy shaped by uncertain demand and rapid market changes, all of these
forecasting methods must be harnessed within one practical, comprehensive solution suite. Multi-variate
demand signal management sorts out the flood of data in a structured way to recognize complex patterns
and to separate actionable demand signals from a sea of “noise”. A best-in-class forecasting system is one
that provides flexibility for users to weight elements and override key parameters in the forecast calculation
based on their intuitive knowledge and market expertise.
Logility’s powerful software solutions help planners leverage the best methods, spot trends and
forecast demand signal changes more quickly, and sense and respond to market changes and inventory
investments and deployments.
To learn how Logility can help you make smarter decisions faster, visit www.logility.com.