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.

Read more in the Supply Chain Magazine’s trend interview (in Dutch) with EyeOn Managing Partner and Senior Consultant André Vriens.

This article might also interest you
How to plan in a post-pandemic world: Five key focal areas in 2021

The post Seize opportunities with the help of digital transformation 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.

New technologies offer the opportunity to accomplish a more data driven supply chain. This made companies rethink their business planning processes. In this webinar we shared our vision on how to embark on the digital transformation journey and how to take the momentum of the Covid situation to take action.  

This webinar is part of the ‘Accelerate the digital transformation’ webinar series in which we touch upon the different building blocks of end-to-end supply chain transformation, with a webinar on the EyeOn approach regarding this topic, the next-generation IBP, and tool selection.  Also we will review a number of the current APS solutions on the market. Check out our upcoming events.

The post Webinar ‘Accelerate the digital transformation – Introduction’ appeared first on EyeOn.

Inventory reduction strategy

Many markets and companies are recovering from the disruptions COVID-19 has caused. Markets are opening up again and production facilities have resumed production. Faced with a high uncertainty on how the economy and markets will recover, CEOs are forced to cut costs and free up money by reducing stock. On the other hand it is increasingly important that customers are served at the desired level, thus a solid stock reduction strategy is required. This brings us to the question of this blog: inventory reductions, what are do’s and don’ts?

Don’ts!

1. Focus on overall lowering safety stocks. I more and more hear companies start to decrease their safety stock levels across all items with a certain percentage, even up to 30%! Don’t do it. It will definitely lead to lower service levels – amplified by the increased demand and supply uncertainty that most companies currently face – and a further increase of the bullwhip in the supply chain, with all its well-known consequences.
2. Set target (safety) stock levels using forecasts from stationary statistical forecast models. Demand during the pandemic is definitely not stationary. What to do instead? See number 4 in the list of do’s!
3. Spend time on modelling the impacts of the epidemic itself, how long recovery will take and if and when a “second wave” can be expected. Leave it to the experts!

Do’s

1. Increase production flexibility/decrease lot sizes. Creating “flow” and reducing lead times in your supply chain by lowering lot sizes will lead to a lower cycle stock, but make sure operational costs only grow marginally!
2. Reduce complexity/diversity product portfolio. Make sure you can deliver your important (A) items and reduce the number of unimportant, less value adding (C) items. A nice example from the pasta industry can be found by visting this page.
3. Reduce stock of slow/non-movers by writing them off, increase sales/marketing activity or stop production. But make sure you identify the real slow/non-movers!
4. Forecasting: extract real business drivers and underlying demand pattern and use them to forecast the “rubber duck” or “bathtub” curve.
5. Create E2E-visibility of stock levels across the supply chain. In multiple projects lately we identified many quick wins: double safety stocks in the supply chain were removed, and/or stock was balanced over locations having under- and overstock.

Need help for a quick start?

Is it that simple, just follow the above guidelines? No, every situation requires a tailored approach and not every “do” might be that effective. As a first step, it is important to know how your “inventory pie” is divided over the different sub inventory types (safety stock, cycle stock, strategic stock, transit stock, etc.) and across the supply chain. Based on that, take actions on the most promising stock type(s) and location(s). A nice example of how an inventory dashboard creates such insights, can found by visiting this page. Please feel free to reach out to us if you would like to discuss what your company needs as a first step, our experts look forward to support you along this challenge!

The post Reduce stock value to recover from Covid-19 turbulences appeared first on EyeOn.

Proposed solution: rubber duck curve

The proposed solution is the rubber duck curve forecast adjustment. The figure below is called the rubber duck curve, because it has the shape of the beloved bath toy. The idea is simple and intuitive. The method divides the time horizon into four moments: regular sales, disruption, recovery, and new normal. The method assumes that the regular sales will follow the patterns of the statistical forecast in a stable environment. Then, the impact of Covid-19 kicks in, causing a decline in demand. A period of disruption starts. As measures to control the infection are eased, the sales are expected to recover. This recovery phase can be perceived in one period, or over a period of time. It is also possible that there is no recovery and the new normal picks up directly from the disruption phase. The new normal is defined as the period with new stable sales. From this point on, the statistical forecasting methods can take over again. Beware that the impact of disruption can also have an opposite effect. Thus, a positive effect on the sales.

Application and implementation

The method is a great concept, but how do you apply it to your forecasting process? To be able to apply this method you need to have knowledge about the expected impact of the virus. This knowledge is represented by three key moments in the curve: start of disruption, start of recovery, and start of the new normal. Each moment is characterized by a starting point in time, and an expected percental change in sales with respect to the forecast before the disruption started. If you have identified each moment, you can impute the expected behavior to your statistical forecast. The statistical forecast incorporates the trend and seasonal effects, so the future forecast has the right demand pattern adjusted with the knowledge about the expected impact.

The way in which these moments are specified is through a template. Information about the expected disruptions is gathered here. The template is flexible to allow the specification of each of the moments at the right hierarchy level. This ensures that the adjustments match the level at which companies can expect the disruption. Additionally, this allows for a fast adjustment of the entire portfolio, without having to provide an expected change at the lowest hierarchy levels.

Example

The figure below shows an example of rubber duck adjustments and a filled-in template. The first columns of the template indicate the hierarchy level which is impacted by Covid-19. Then, the next three columns indicate ‘when’ a new phase starts, and the last three columns indicate what the percental change is.  The graph shows that starting on May 2020 there is a change of -30% in the expected demand; the disruption phase has started. Then, in November 2021 the recovery period starts with a change of 15% in comparison to the reference forecast. Finally, the new normal starts on February 2021 and the sales are expected to stabilize again. From this moment on, the statistical forecasting method can take over again and rubber duck adjustments are no longer needed.

Benefits

Adding the rubber duck curve to the statistical forecast can provide a good foundation for future demand planning. Consequently, there is no need to adapt each individual forecast in an enrichment tool manually due to unrealistic mid/long term statistical forecast figures. Furthermore, the template can be updated easily during each forecast cycle based on the latest insights. Lastly, the rubber duck curve can be implemented fast and can be easily reapplied in case of a another Covid-19 wave. 

In conclusion, combining the power of statistical forecasting with the knowledge about expected impact on your business as a rubber duck curve, provides you with a Covid-19-proof method for forecasting.

Please feel free to contact us for more support!

The post Mid and long-term demand forecasting in times of COVID-19 appeared first on EyeOn.

How to navigate your recovery?

As markets are opening up again, it is now time to choose and implement your recovery strategy. Underlying questions include: when will recovery start? How large is the initial demand leap and over which period of time is it spread out? What will be the ‘new normal’? How will the lockdown be ‘unlocked’? Can we expect a second or even more lockdowns in the future? For companies which experience the opposite effects of Covid-19: how to bring production rates back to the ‘new normal’? A lot of unanswered questions, but that doesn’t mean it is impossible to prepare for recovery.

Admittedly, every supply chain and company experiences a different challenge and requests for a tailored approach. Yet, we believe the following three ingredients are key for a successful recovery:

1. Create end-to-end supply chain visibility, in order to determine (rough) estimates of the demand and supply recovery parameters.
2. Apply scenario planning techniques: using rough estimates, what-if simulations help you gain insight in and explore what variables dominate future stock developments across the supply chain. The insights also create focus on what variables deserve additional forecasting attention.

3. Team up with your supply chain partners! Align your recovery plans with your supply chain partners (or even better: create them together) to ensure its feasibility.

Supply chain recovery: the bumpy road ahead of us

The challenge many supply chains face reminds of the song by City to City: the road ahead for your supply chain is bumpy, with miles of the unknown ahead of you. Nothing is sure along the way, not even tomorrow. Supply chain scenario planning can help you navigate out of the lockdowns, not only to select the initial route but also to change it along the way. We demonstrated our scenario planning approach with a concrete example during a live webinar on May 13^th. Our presentation can be found here.

Our experts look forward to support you along this challenge, so please reach out to us and join forces. For the rest: Take life the way it comes. And above all, take care of yourself and your families!

The post Supply chain scenario planning: navigate out of COVID-19 lockdowns appeared first on EyeOn.

The vast majority of companies rely on statistical forecast models to predict their demand. These models use time series of historical sales to estimate future demand. A key assumption of these models is that the underlying system remains unchanged. In other words; the relationship between historical and future demand does not change. Clearly this assumption is violated during the current Corona crisis and as a result we should be very reluctant to rely on statistical models. The current circumstances present both a need and an opportunity to predict demand more dynamically. This requires different information than just historical sales; it requires tapping into other data sources that provide insights regarding future demand. One of these data sources to consider are the open orders from your order book.

What exactly are open orders?

Open orders refer to demand that has been placed in advance. In other words, future demand that is already known at present. Although many companies have access to open orders via their order book, this information rarely finds its way into the statistical forecast. At best, planners use this information manually to adjust the statistical forecast. It would be more efficient, however, if the statistical forecast could somehow already take this information into account. Sure, that sounds good, but how can we do this? Well, 18th century clergyman and mathematician Thomas Bayes provides us with an interesting toolkit to take on this problem.

Bayesian inference

Before we jump right into it, let’s take a moment to discuss the basics of Bayesian inference. Bayes’ Theorem provides us with a quantitative framework for updating our beliefs as the facts around us change or new information arises. This framework is captured in his famous equation:

Although the formula might look intimidating at first, what it essentially boils down to is this: whenever we receive new information, how much should we let it affect what we currently believe to be true? Does the new information support the initial belief, dispute it, or not affect it all? Are you starting to see how this method might help in incorporating order book data into the statistical forecast?

Applying the Bayesian line of reasoning to our forecast problem

The statistical forecast provides us with the initial belief regarding what the demand will be in the future. By contemplating the order book, however, we receive new information that will either support or dispute this initial belief. Using Bayes’ equation, we can then calculate the Posterior distribution. This distribution represents our belief regarding future demand after taking into account both the initial statistical forecast and the open orders. 

Example

Let’s clarify with an example. Imagine that your (initial) statistical forecast for one month ahead is 24. Moreover, your order book shows that, at present, you have 6 pre-orders in the system for one month ahead. Finally, the last piece of information that we need is an estimate of the probability that an order for next month is already known at present. Assume that this probability is equal to 50%. The question now becomes ‘How to use the number of pre-orders to adjust the initial statistical forecast?’. 

1. The first step is to fit a distribution with mean equal to the initial forecast (in our example 24). This distribution (in blue) represents our initial belief about what the demand for next month is going to be.
2. Next, we use Bayes’ rule to adjust the initial forecast. To put it plainly, in this example the order book tells us that future demand is likely to be lower than we initially expected based on the statistical forecast. Hence, the Bayesian forecast incorporates this information and lowers the statistical forecast (from 24 to 19). 

Final thoughts 

As a consequence of the Corona crisis, it is possible that your market is currently facing drops in demand, increases in demand, or significant demand shifts over time or between products. In any case, your order book might provide valuable information, signalling where your future demand is heading. With the Bayesian toolkit, you can incorporate this information automatically without having to check every order yourself.

Eager to learn more about this topic or curious if this method could also be valuable for your organisation? Please do reach out to Rijk van der Meulen.

The post Dynamic demand prediction in times of Covid-19 appeared first on EyeOn.

This was one of the questions that a group of executives from FMCG, food and retail companies discussed on the EyeOn inspiration event in the ADAM tower in Amsterdam Wednesday 13^th of November.

For many companies’ market circumstances are volatile, varying from global politics to individual consumers who can shift buying preferences ‘in a click’. Next to that data is everywhere, the war on talent is increasingly challenging and sustainability is driving major changes in how to be successful. Although many value chains are more transparent than ever before, this hasn’t made us better planners or decision makers (yet).

Luckily the possibility of next level planning and decision processes is available. Discussing the possibilities of no -touch planning with self-driving cars, the clear conclusion by all was that we need to improve 1-2 levels in the next few years.

Tooling and data are no limiting factors anymore. The group agreed that organisational readiness, cultural change and embracing the potential of analytics deserve the most attention in making the next steps.

The changing world of internet platforms is making clear that business models are changing and most likely will impact everybody not only in B2C but also in B2B, according to Maurice Jongerius from Bol.com. The consequences are not only technology driven but even more on the people side. Leaders and managers should start changing themselves!

Conclusion: Transformation thinking is key!

Overseeing Amsterdam, the conclusion was that transformation thinking is key!
5 best practices to start transforming tomorrow were shared by André Vriens & Jan Veerman:

• Ask the right questions for your business!
• Start collecting and storing data as of tomorrow
• Build strong analytical skills and embed analytics in business processes
• Collaborate with all relevant stakeholders in and outside your company
• Remain agile, just start!

Do you want to know more? Please contact André Vriens to discuss!

The post How should business leaders cope with challenging environments? appeared first on EyeOn.