Moving holidays are holidays that occur each year, but where the exact timing shifts from the perspective of the Gregorian calendar system. Chinese New Year (CNY) is an example of a moving holiday, it is based on the lunar calendar. Chinese New Year most often falls in February but can also occur in January. Since the date of Chinese New Year changes from year to year, the effect of this holiday can impact sales in multiple months. It is often the case that production accelerates some time before the start of Chinese New Year, almost completely stops during the holidays, and finally rises to the regular level after the holidays. In these cases, the effect of the holiday is not confined to the seasonal component of the time series since the seasonality rhythm (based on the lunar calendar) is not in line with the demand forecast rhythm (based on the Gregorian calendar). This often leads to a significant decrease in the performance of the statistical forecast (i.e. lower forecasting accuracy and higher bias) for the months affected by the holiday.

Combining statistical forecasting with machine learning for increased forecasting accuracy

Conventional statistical models (e.g. moving average and exponential smoothing) are widely used within the industry to predict demand. Often with good reason, since these models usually perform reasonably well and they are intuitive and easy to interpret for planners. We have seen, that statistical models (even if we add a seasonal component) are not able to model the complex effects of moving holidays. Based on these observations, we designed a new approach using machine learning to predict ‘uplift’ factors that are used to scale the baseline forecasts. We give a detailed description on how EyeOn addresses this challenge in this blog post: How to make better demand forecasts for Chinese New Year using machine learning

Results case study previous blog

Going beyond Chinese New Year

The results of modeling Chinese New Year for a large multinational are very promising. In this case study, we find enormous improvements in forecast bias. For ten different regions, we achieve an average absolute bias reduction of 68 percent points. We don’t stop there, however! The next step is to extend the logic on two fronts: (1) automatic detection of significant events (rather than manual) and (2) applying the logic to a wider variety of moving holidays in addition to Chinese New Year.

First, we implement automated event detection to determine whether an event significantly impacts a time series in a certain period. If so, we use our machine learning model to apply a correction to the baseline forecast. If not, we just use our statistical baseline forecast. The event detection consists of three statistical tests, which determine if the sales during an event are significantly impacted. The identified events are corrected by predicting uplift factors for the event periods. Automatic event detection results in less manual intervention from the forecaster and provides statistical proof that an event has an impact on the time series. Making you less dependent on the forecaster’s ability to interpret and their experience of whether a correction should be applied.

Second, besides the uplift factors, the event sales are corrected to improve the forecast performance in the periods after the event. We do this correction to avoid the sales disruption during the event impacting the statistical baseline forecast. In the case of Chinese New Year, the correction is used to make sure that the statistical forecast level will not be too low after having three negatively impacted sales months.

The extension of this logic makes it possible to better forecast moving holidays and to rely on statistics to determine which events should be corrected. Making it applicable for more customers and events, resulting in improved forecasts during moving holidays. We applied the new extended logic to two other events besides Chinese New Year in a new case study. Comparing the results of the developed method to the results of conventional time series models (e.g. moving average/simple exponential smoothing). In this case, we find an average increase of 13.1 percentage points in forecasting accuracy and a reduction of 18.7 percentage points in the bias for the affected months. Note that for the non-impacted months, the two forecasts are identical.

Challenge accepted, challenge completed: increased forecasting accuracy

Moving holidays can have a significant impact on demand. Moreover, modeling this effect can be challenging since the effect of the event impacts a different period (e.g. week or month) each year. We accepted this challenge and developed a new approach where the conventional time series forecast is complemented with a machine learning algorithm that models the effect of moving holidays. This approach is now available for a large variety of events and customers.

Eager to learn more about this topic or curious if this method could also be valuable for your organization? Please reach out to us!

The post Increased forecasting accuracy during moving holidays appeared first on EyeOn.

Turn your challenge into an opportunity for future success. Our experts recommend the following steps to accelerate supply chain digitalization and end-to-end transformation of your supply chain:

Start supply chain digitalization from a solid planning foundation

To realize robust end-to-end transformations it is key to start from a solid supply chain foundation. Start by developing an optimal planning model that translates strategy into operational planning principles. Understand the value drivers in your supply chain and define well connected planning processes and a matching organizational setup to facilitate high value planning decisions. Your process re-design needs to be focused on E2E connections, linking the different layers of planning and control of your supply chain through smart alerts.

Focus on high-quality data and planning parameter automation

A holistic planning model and integrated planning process require robust and well-integrated data structures to drive execution. A well-defined data model and supportive governance processes across your IT landscape are vital to secure setup, maintenance, and right-sizing of your planning parameters. It all starts with having a complete, accurate, and well-aligned master data fundament. Building on this solid baseline you can have the data work for you to build a process for smart-touch automation of your critical planning parameters. As your supply chain develops you can real time update your vital parameters to improve overall planning accuracy.

The full value of advanced planning solutions

A new generation of planning systems promise embedded cognitive automation to make real-time recommendations, predict outcomes, and take timely supply chain decisions. It is crucial that your functional design supports the E2E planning process, connects well to your enterprise-wide data model and allows progressive digital innovations (such as AI and advanced analytics). To realize the full value of your APS (advanced planning and scheduling) system you need to build the right in-house capabilities. Your planners require the right data processing, analytical, and data science skills to operationalize and embed digitalization into the day-to-day planning routine.

Building a digital ecosystem

If you are ready for the next step, look beyond traditional planning systems and investigate how systems of innovation can help to boost planning capabilities in a digital ecosystem. Data science platforms like Dataiku can bring advanced analytics capabilities in-house quickly and can be plugged into the systems that are already in place, like ERPs (enterprise resource planning), APSs, or data lakes. These systems of innovation allow data scientists to make the most out of their expertise, allow for easy collaboration between them and the business, and enable you to go from prototype to production quickly but robustly. Building an ecosystem of digital capabilities will help your organization to accelerate the digital transformation through the implementation of focused innovations.

We are here to support you in your supply chain digitalization journey!

Are you looking for ways to accelerate the digitalization of your E2E supply chain planning process and are you in need of the right expertise and hands-on support to make it happen? We are here to help you in this journey. Together we transform, automate, and optimize. Learn more about E2E transformations or contact us now!

This is the time to build resilient, agile, and sustainable supply chains, maximizing the benefits of digitalization and advanced analytics.

The post Accelerate the digitalization of your supply chain appeared first on EyeOn.

Marco van Alfen, IBP expert: “I know examples of companies that have purchased very expensive software for forecasting, but in the end do not dare to trust what the system prescribes. It doesn’t happen by itself. If you want to do this well, you have to invest in data, people, systems and processes.”

Learn the four steps to create a data-driven culture in Alex van Groningen’s article (in Dutch) with Marco van Alfen, Sr. Business Consultant and IBP expert at EyeOn.

Read more about forecasting and planning!

The post More accurate forecasting: Invest in data-driven culture! appeared first on EyeOn.

Ask our experts about the EyeOn approach for driver-based forecasting: Contact us!

The post Video: Driver-based forecasting with machine learning appeared first on EyeOn.

Data latency is one of the biggest issues in supply chain decision-making. Too much time is lost from the moment an event happens until data becomes available (capture latency), until it has been analyzed (analysis latency), or until a decision has finally been taken (decision latency). Due to this loss of time from the occurrence of the event to acting on it, the value and impact of that decision have dramatically decreased.

We don’t just need to make the right decision, we need to make that decision really quickly. For this purpose EyeOn is developing smart-touch planning: with better technology we enable you to make faster decisions, thereby increasing the impact.

How smart-touch planning enables fast and informed decision-making in supply chain

We help you reach maximum quality data with minimum effort, provide a user-friendly ecosystem integrated with your own infrastructure, and enable your planners with best practices and AI insights for fast, informed supply chain decision-making. Smart-touch planning automates obvious decisions, makes recommendations where additional information is available, and flags special cases where human intervention is necessary.

Smart-touch planning is adapted to your company. It means making planning as simple as possible; adding complexity where needed to ensure the best possible forecast for your company and industry. Explainability is everything, to make sure that you get the most out of it. We focus on user-centricity and integrate smart-touch planning into your existing workflow.

This is what Smart-touch planning could look like:

An AI gives hints to users, enabling planners to make high-quality decisions.

Data engineering – the basis of future data analysis

We bring together data from all sources and connect it so that it forms the basis of future data analysis. We offer training to help you in collecting and connecting high-quality data and we provide standard templates to ease collecting the right data. Our Honeycomb software organizes data and provides insights & dashboards on data quality.

Better forecasting through feature engineering 

We take raw data and turn it into features that can be used for better supply chain decision-making. We combine data science techniques and existing libraries for feature generation with specific industry and customer knowledge to extract the most information. More useful features will give a better, more explainable, higher accuracy, and lower bias forecast. Using standard analyses and dashboards we identify which factors impact your business, giving you an even better understanding of your business dynamics and e.g. promotion effects.

Model ecosystem for effortless use

We are building a model ecosystem that provides a scalable infrastructure and set of best-of-breed models to predict, optimize, and explain. Both classic models (e.g. Holt-Winters) and more advanced models (Prophet, Catboost, ..) are available. Automated pre-configuration is included: the need to tune models is minimized – the aim is to enable everyone to use all models, effortlessly. This gets us great results quickly and gives the data scientists more time for in-depth analysis of individual cases. 

Enable your planners to be more effective

We have experience and unique technologies to give planner behavior feedback that enables you to increase the value provided by the planners. The goal is for planners and your company to benefit maximally from our smart-touch planning approach. We share best practices and standard insights to enhance planner behavior and decisions through training and technology.
We are currently productizing these insights, transforming dashboards and insights from our consultants to automated AI-based feedback.

Let’s discuss how smart-touch planning can improve supply chain decision-making, and make it easier and faster: Contact one of our experts! 

The post Improving supply chain decision-making: make better decisions, faster appeared first on EyeOn.

Moving holidays are holidays that occur each year, but where the exact timing shifts from the perspective of the Gregorian calendar system. Examples of moving holidays include Easter and Chinese New Year (CNY). Easter generally falls in April but can also fall in late March. Chinese New Year mostly falls in February but can also occur in January. Since the date of these holidays changes from year to year, their effect can impact two or more months depending on the date. Related to Chinese New Year, for example, it is often the case that production accelerates some time before the start of Chinese New Year, almost completely stops during the holidays, and finally rises to the regular level after the holidays. In these cases, we don’t talk about seasonality in forecasting, because the effect of the holiday is not confined to the seasonal component of the time series since the seasonality rhythm (based on the lunar calendar) is not in line with the demand forecast rhythm (based on the Gregorian calendar). This often leads to a significant decrease in the performance of the statistical forecast (i.e. lower accuracy and higher bias) for the months affected by the holiday. In this blog, we will explore how EyeOn addresses this forecasting seasonality challenge for one of our customers.

Combining conventional statistical forecasting with machine learning

Conventional statistical models (e.g. moving average and exponential smoothing) are widely used within the industry to predict demand. Often with good reason, since these models usually perform reasonably well and are intuitive and easy to interpret for planners. We have seen, however, that these models (even if we add a seasonality component) are not able to model the complex effects of moving holidays. Based on these observations, we designed the following approach at EyeOn:

1. Start with a conventional statistical model to obtain a baseline forecast.
2. Predict “uplift” factors for months affected by the moving holiday(s). In other words, in this step, we are estimating how much higher or lower the demand is in a given month, relative to the baseline demand. Note that the uplift factor can also be smaller than 1. In that case, we are actually scaling down the baseline forecast.
3. Multiply the baseline forecast with the uplift factors to obtain the final forecast.

Obviously, this approach relies to a large extent on how well we are able to estimate the uplift factors, and for this step, we are harvesting the power of machine learning. In the context of modeling moving holidays effects, a well-known type of regressor called the Bell-Hillmer interval, has proven to be very useful. Assuming that the holiday effect is the same for each day of the interval over which the regressor is nonzero in a given year, the value of the regressor in a given month is the proportion of this interval that falls in the month. Using this logic, we can thus define multiple intervals to model the backward and forward effect of a moving holiday. These Bell-Hillmer regressors are used as features in a machine-learning algorithm that uses gradient boosting on decision trees.
Although the above description might sound daunting at first, what it essentially boils down to is this: based on the characteristics of one or more moving holidays, we let a smart algorithm learn by which factor we need to adjust the statistical baseline forecast.

Case study on forecasting seasonality: modeling the impact of Chinese New Year for a large multinational

Chinese New Year is China’s most important holiday and the largest annual mass migration on the planet. Since most elderly parents live in rural villages and their children work in the cities, the “chunyun” (spring migration) creates approximately two to four weeks of radio silence from the entire country, including your suppliers, contract manufacturers, and partners. During this time, almost everything shuts down.
All of this poses serious complex supply chain planning challenges for all companies operating in Asia. The graph below shows one of these challenges. The vertical bars represent the sales quantity per month from January 2017 up until March 2021. The orange-shaded bars annotate the months December, January, and February where the effect of CNY is clearly visible. Also, note that the effect varies considerably from year to year. For instance, in the years on which CNY fell in January (2017 and 2020), the sales in January were impacted significantly more compared to the years when CNY fell in February. The blue line in the graph depicts the forecast generated by conventional time series models (i.e. moving average/simple exponential smoothing). Note that, although we supplemented these models with a seasonal component, they do not fully capture the effect of CNY. This results in reduced forecast accuracy and greatly increased bias for the months affected by CNY.
The green line shows the forecast as generated by the method proposed in this blog. Already from looking at the graph, it becomes clear that this forecast outperforms the conventional time series forecast. In this case, we found an increase of 3.5 percentage points in forecast accuracy and a reduction of 44.6 percentage points in the bias. Note that for the non-impacted months, the two forecasts are identical.

Final thoughts on forecasting seasonality

While conventional time series models have a good track record and are the de facto standard in the industry, they are often not equipped to capture more complex effects, such as moving holidays. If these effects are large (such as in our example with Chinese New Year), this could lead to diminishing performance of the statistical forecast. To address this issue, we implement this new forecast approach where the conventional time series forecast is complemented with a machine learning algorithm that models the effect of moving holidays.

Eager to learn more about this topic or curious if this method could also be valuable for your organization? Please reach out to us: Rijk van der Meulen, Anne de Vries, or Dan Roozemond.

The post Improve forecasting seasonality for Chinese New Year using machine learning appeared first on EyeOn.

No-touch planning, automated generation of master data, robotic process automation. On the one hand advanced automation seems and is closer than ever.

Advances in general purpose AI (think DeepMind) or focused AI (voice assistants, photo pattern recognition) make it easy to imagine the application in supply chain planning is only a matter of time. In this light, episode 47of the ‘After On’ podcast is worthwhile the listen.

Still, we struggle on a daily basis with bad master data, erroneous or time-consuming Excel files, difficult to interpret solvers and cumbersome scenario planning. Maybe an automated supply chain is just one leap to far?

Take a step back and think about the technology we use every day. What’s the last time you’ve used physical maps to navigate? Go to the library? Remember the stock traders shouting on the exchange floor? Technology slowly advances and seamlessly intertwines with daily life. 

The same holds for applications in supply chain management. We just executed a deep dive in the level of automation we apply in analyses we execute for and in applications delivered at our customers. The degree of automation in a typical project has risen to a level we would not find say 10 years ago. 

For example, more and more steps of setting up a forecasting process are automated. Manual segmentation is not required anymore. Segmentation is automated, just like many of the steps to select the best forecast approach for each category.

 

 

In our work to reduce data latency through speeding up requirements propagation, demand trend recognition and shorten planning cycles, we see ever more opportunities to automate. Take a multi node divergent supply chain with many routing choices per node, with varying lead times and yields. A detailed understanding of critical starting material requirements requires a full plan propagation, with a latency of (at least) one planning cycle. An automated ad hoc propagation using predicted lead times and material requirements – learning from past routing decisions and outcomes – delivers predicted raw material requirements with accuracy fit for planning on demand, rather than once every cycle at best.

 

 

And it can work the other way around. An analysis of supplier performance giving insight of performance against planned parameter, triggered the development of an automated update and learning loop feeding supplier management, QC lab priorities, master data management and planning.

 

 

 
Automation based learning loop for supplier management 

Any step, no matter how small, brings the no touch supply chain a little closer. At minimum it provides answers to the questions asked. In many cases, though, it triggers efforts to significantly reduce data latency and implement true automated learning loops.

Want to learn more? Ready to take the first step and implement? Feel free to reach us through our website or read the document “Data Engineering” to familiarize with the approach EyeOn takes on making improvements in this area.

In striving for success, large companies have to continuously struggle against growing internal complexity. EyeOn helps our clients manage this complexity by designing, implementing and executing excellent planning processes as a discriminating factor for this success. In order to achieve this, we develop and share knowledge about top level planning and forecasting, with constantly demonstrable return on investment for our clients.

The post Every journey starts with a first step, and before you know it you’re running appeared first on EyeOn.