This website does not support Internet Explorer. Please, switch to Edge, Chrome, or Firefox browser to view this page.

Learn about IE support
Best Practices

Improving forecast accuracy with AI and ML techniques

Improving ml tehcniques blog header
Published: Reading time: 5 min
o9 Solutions The Digital Brain Platform
o9 SolutionsThe Digital Brain Platform
Published:

The current state of supply chain

Today’s supply chains are being disrupted like never before. Consumers demand the best product selection, rapid delivery, and customizability options at the best prices. And all of this while the world faces uncertainties brought about by climate change, trade tensions, resource scarcity, the effects of COVID-19, and most recently, the energy, food, and finance crisis caused by the war in Ukraine.

Forecasting is a critical activity that helps companies predict future demand, mitigate potential risks, and capitalize on emerging opportunities. Still, the current volatile environment is increasing pressure on businesses to reduce value leakage. Organizations face problems because traditional forecasting methods, siloed processes, and legacy technologies make value leakage extremely difficult to prevent. Therefore, businesses are urgently focusing on digitally evolving their forecasting capabilities and operations.

Next-generation technologies such as artificial intelligence (AI), specifically machine learning (ML), can significantly increase a business’s forecast accuracy, help it navigate a volatile demand landscape, and ensure its continued growth. A few years ago, an AI/ML-based planning solution would have been a maturity-phased objective, but now it has become a business imperative to reduce value leakage.

The limitations of traditional forecasting

Traditional forecasting methods are limited as they mainly use historical data to predict future demand. While this approach may have worked in the past when demand was relatively stable, rapidly fluctuating demand means that companies need to shift toward utilizing the wealth of real-time, external data about the market to create more accurate forecasts.

However, most companies are not leveraging these external drivers of demand because they are still using the traditional forecasting approach based on historical data. Legacy forecasting systems often cannot ingest or use the leading indicators or demand drivers that can unlock faster, more accurate forecasts and provide valuable insights.

The emergence of leading indicators of demand

Leading indicators of demand are data types that have predictive values or ‘drivers’ for forecasting and are becoming crucial in obtaining increased forecast accuracy. Broadly speaking, there are two categories of demand drivers:

Market knowledge includes drivers such as consumer demographics, Gross Domestic Product (GDP), and interest rates.

External data, such as the Internet of Things (IoT), social media, review sites, competitor websites, and the news, all help create significantly more comprehensive forecasts.

Depending on the industry and business, real-time external data such as local weather, events around the stores, road conditions and traffic, click-views, and search trends on the Internet can add even more refinement to a forecast.

The sensing and shaping of demand

But to make meaningful inferences from these demand drivers, the data must be converted into actionable knowledge to improve forecasting accuracy and decision-making. In the short-term horizon, this is known as demand sensing, where drivers in the immediate future are evaluated for fast business reaction and impact.

Demand Sensing needs to be done promptly to allow businesses to capitalize on the insights. Dispensing with time-series forecasting and embracing intelligent forecasting through AI techniques such as ML forecasting can enable demand sensing to take place automatically, providing a decisive competitive advantage.While AI/ML solutions help create faster and more accurate statistical forecasts, they can also improve decision-making related to commercial activities. These activities are called demand shaping, where machine learning can sense, analyze and recommend actions to the planners.

Examples here are pricing, promotion, new product, and event initiatives that can be set to achieve desired business objectives, such as optimizing revenue potential, increasing margin, or maximizing sales volume.

The future of forecasting with machine learning (ML)

Furthermore, these next-generation technologies can take leading indicator data and create a forecast free of human bias. They can constantly learn which leading indicator data best predicts demand for more accurate predictions, down to the levels of location, item, and time.

A deeper understanding of how these drivers influence demand combined with the learning abilities of AI and ML forecasting enables a higher degree of automation of demand forecasting, freeing up planners to address exceptions and more complex cases. Here are some industry examples where AI/ML and demand drivers can help organizations manage value leakage:
For the chemical industry, the ability to link macroeconomic data at the sector and country level allows for accurately identifying different demand streams, thus creating a composite forecast across horizons for contracting and tenders.

For the food industry, connecting leading demand indicators, such as weather forecasts and satellite images, and utilizing ML algorithms to recognize patterns, outliers, and seasonality to predict optimal harvest time creates yield and supply chain efficiencies. As a result, planner productivity gains of up to 60% have been reported.

For the automotive industry, leading indicators such as interest rates, inflation, and disposable income can improve forecasting for new car sales and leasing. Similarly, vehicle component usage data can be collected to refine the prediction of failure rates and assist the management of maintenance servicing and the related inventory levels of spare parts.
Ultimately, automated intelligent ML forecasts do not just increase productivity; they produce plans optimized to the degree that neither manual nor traditional solutions can deliver.
If you are interested in learning more, read our whitepaper, which discusses how AI techniques like ML forecasting can significantly improve forecasting accuracy, optimize how you plan for demand, and evolve your company’s DNA from traditional to digital.

Leaders gartner banner image

Gartner® Magic Quadrant™ 2023

o9 Solutions recognized as a Leader in the 2023 Gartner® Magic Quadrant™ for Supply Chain Planning Solutions. Download for free the full Magic Quadrant.

I agree to the privacy policy.

About the author

o9 Solutions The Digital Brain Platform

o9 Solutions

The Digital Brain Platform

o9 Solutions is a leading AI-powered platform for integrated business planning and decision-making for the enterprise. Whether it is driving demand, aligning demand and supply, or optimizing commercial initiatives, any planning process can be made faster and smarter with o9’s AI-powered digital solutions. o9 brings together technology innovations—such as graph-based enterprise modeling, big data analytics, advanced algorithms for scenario planning, collaborative portals, easy-to-use interfaces and cloud-based delivery—into one platform.

Resources

View our related articles, white papers, use cases & videos

article4 min

How to Use Your Data Lake for Storage

by Stijn-Pieter van Houten
O9 mof blog how to use your datalake for storage header
article12 min

Retail Analytics Explained

by o9 Solutions
O9 blog what is retail analytics header
Guide

10 Steps for Setting Up an Effective Transformative IBP Process

O9 ibp cheatsheet mockup 3
article16 min

Day 1 Recap: aim10x digital 2024

by o9 Solutions
Day 1 recap web banner 920x1080px
video

The Aftermovie: aim10x live Singapore

2 min
The aftermovie: aim10x live singapore thumbnail
video

Why the Future of CPG Supply Chains is Autonomous

52 min
Why the future of cpg supply chains is autonomous thumbnail