What to do when a statistical forecast doesn’t make sense

Sometimes a statistical forecast just doesn’t make sense.  Every forecaster has been there.  They may double-check that the data was input correctly or review the model settings but are still left scratching their head over why the forecast looks very unlike the demand history.   When the occasional forecast doesn’t make sense, it can erode confidence in the entire statistical forecasting process.

This blog will help a layman understand what the Smart statistical models are and how they are chosen automatically.  It will address how that choice sometimes fails, how you can know if it did, and what you can do to ensure that the forecasts can always be justified.  It’s important to know to expect, and how to catch the exceptions so you can rely on your forecasting system.

 

How methods are chosen automatically

The criteria to automatically choose one statistical method out of a set is based on which method came closest to correctly predicting held-out history.  Earlier history is passed to each method and the result is compared to actuals to find the one that came closest overall.  That automatically chosen method is then fed all the history to produce the forecast. Check out this blog to learn more about the model selection https://smartcorp.com/uncategorized/statistical-forecasting-how-automatic-method-selection-works/

For most time series, this process can capture trends, seasonality, and average volume accurately. But sometimes a chosen method comes mathematically closest to predicting the held-out history but doesn’t project it forward in a way that makes sense.  That means the system selected method isn’t best and for some “hard to forecast”

 

Hard to forecast items

Hard to forecast items may have large, unpredictable spikes in demand, or typically no demand but random irregular blips, or unusual recent activity.  Noise in the data sometimes randomly wanders up or down, and the automated best-pick method might forecast a runaway trend or a grind into zero.  It will do worse than common sense and in a small percentage of any reasonably varied group of items.  So, you will need to identify these cases and respond by overriding the forecast or changing the forecast inputs.

 

How to find the exceptions

Best practice is to filter or sort the forecasted items to identify those where the sum of the forecast over the next year is significantly different than the corresponding history last year.  The forecast sum may be much lower than the history or vice versa.  Use supplied metrics to identify these items; then you can choose to apply overrides to the forecast or modify the forecast settings.

 

How to fix the exceptions

Often when the forecast seems odd, an averaging method, like Single Exponential Smoothing or even a simple average using Freestyle, will produce a more reasonable forecast.  If trend is possibly valid, you can remove only seasonal methods to avoid a falsely seasonal result.  Or do the opposite and use only seasonal methods if seasonality is expected but wasn’t projected in the default forecast.  You can use the what-if features to create any number of forecasts, evaluate & compare, and continue to fine tune the settings until you are comfortable with the forecast.

Cleaning the history, with or without changing the automatic method selection, is also effective at producing reasonable forecasts. You can embed forecast parameters to reduce the amount of history used to forecast those items or the number of periods passed into the algorithm so earlier, outdated history is no longer considered.  You can edit spikes or drops in the demand history that are known anomalies so they don’t influence the outcome.  You can also work with the Smart team to implement automatic outlier detection and removal so that data prior to being forecasted is already cleansed of these anomalies.

If the demand is truly intermittent, it is going to be nearly impossible to forecast “accurately” per period. If a level-loading average is not acceptable, handling the item by setting inventory policy with a lead time forecast can be effective.  Alternatively, you may choose to use “same as last year” models which while not prone to accuracy will be generally accepted by the business given the alternatives forecasts.

Finally, if the item was introduced so recently that the algorithms do not have enough input to accurately forecast, a simple average or manual forecast may be best.  You can identify new items by filtering on the number of historical periods.

 

Manual selection of methods

Once you have identified rows where the forecast doesn’t make sense to the human eye, you can choose a smaller subset of all methods to allow into the forecast run and compare to history.  Smart will allow you to use a restricted set of methods just for one forecast run or embed the restricted set to use for all forecast runs going forward. Different methods will project the history into the future in different ways.  Having a sense of how each works will help you choose which to allow.

 

Rely on your forecasting tool

The more you use Smart period over period to embed your decisions about how to forecast and what historical data to consider, the less often you will face exceptions as described in this blog.  Entering forecast parameters is a manageable task when starting with critical or high impact items.  Even if you don’t embed any manual decisions on forecast methods, the forecast re-runs every period with new data. So, an item with an odd result today can become easily forecastable in time.

 

 

Service Level Driven Planning for Service Parts Businesses in the Dynamics 365 space

Service-Level-Driven Service Parts Planning for Microsoft Dynamics BC or F&SC is a four-step process that extends beyond simplified forecasting and rule-of-thumb safety stocks. It provides service parts planners with data-driven, risk-adjusted decision support.

 

The math to determine this level of planning simply does not exist in D365 functionality.  It requires math and AI that passes thousands of times through calculations for each part and part center (locations).  Math and AI like this are unique to Smart.  To understand more, please read on. 

 

Step 1. Ensure that all stakeholders agree on the metrics that matter. 

All participants in the service parts inventory planning process must agree on the definitions and what metrics matter most to the organization. Service Levels detail the percentage of time you can completely satisfy required usage without stocking out. Fill Rates detail the percentage of the requested usage that is immediately filled from stock. (To learn more about the differences between service levels and fill rate, watch this 4-minute lesson here.) Availability details the percentage of active spare parts with an on-hand inventory of at least one unit. Holding costs are the annualized costs of holding stock accounting for obsolescence, taxes, interest, warehousing, and other expenses. Shortage costs are the cost of running out of stock, including vehicle/equipment downtime, expedites, lost sales, and more. Ordering costs are the costs associated with placing and receiving replenishment orders.

 

Step 2. Benchmark historical and predicted current service level performance.

All participants in the service parts inventory planning process must hold a common understanding of predicted future service levels, fill rates, and costs and their implications for your service parts operations. It is critical to measure both historical Key Performance Indicators (KPIs) and their predictive equivalents, Key Performance Predictions (KPPs).  Leveraging modern software, you can benchmark past performance and leverage probabilistic forecasting methods to simulate future performance.  Virtually every Demand Planning solution stops here.  Smart goes further by stress-testing your current inventory stocking policies against all plausible future demand scenarios.  It is these thousands of calculations that build our KPPs.  The accuracy of this improves D365’s ability to balance the costs of holding too much with the costs of not having enough. You will know ahead of time how current and proposed stocking policies are likely to perform.

 

Step 3. Agree on targeted service levels for each spare part and take proactive corrective action when targets are predicted to miss. 

Parts planners, supply chain leadership, and the mechanical/maintenance teams should agree on the desired service level targets with a full understanding of the tradeoffs between stockout risk and inventory cost.  A call out here is that our D365 customers are almost always stunned by the stocking levels difference between 100% and 99.5% availability.   With the logic for nearly 10,000 scenarios that half a percent outage is almost never hit.   You achieve full stocking policy with much lower costs.   You find the parts that are understocked and correct those.  The balancing point is often a 7-12% reduction in inventory costs. 

This leveraging of what-if scenarios in our parts planning software gives management and buyers the ability to easily compare alternative stocking policies and identify those that best meet business objectives.  For some parts, a small stock out is okay.  For others, we need that 99.5% parts availability.  Once these limits are agreed upon, we use the Power of D365 to optimize inventory using D365 core ERP as it should be.   The planning is automatically uploaded to engage Dynamics with modified reorder points, safety stock levels, and/or Min/Max parameters.  This supports a single Enterprise center point, and people are not using multiple systems for their daily parts management and purchasing.

 

Step 4. Make it so and keep it so. 

Empower the planning team with the knowledge and tools it needs to ensure that you strike agreed-upon balance between service levels and costs.  This is critical and important.  Using Dynamics F&SC or BC to execute your ERP transactions is also important.  These two Dynamics ERPs have the highest level of new ERP growth on the planet.  Using them as they are intended to be used makes sense.   Filling the white space for the math and AI calculations for Maintenance and Parts management also makes sense.  This requires a more complex and targeted solution to help.  Smart Software Inventory Optimization for EAM and Dynamics ERPs holds the answer.    

Remember: Recalibration of your service parts inventory policy is preventive maintenance against both stockouts and excess stock.  It helps costs, frees capital for other uses, and supports best practices for your team. 

 

Extend Microsoft 365 F&SC and AX with Smart IP&O

To see a recording of the Microsoft Dynamics Communities Webinar showcasing Smart IP&O, register here:

https://smartcorp.com/inventory-planning-with-microsoft-365-fsc-and-ax/

 

 

 

 

Statistical Forecasting: How Automatic method selection works in Smart IP&O

Smart IP&O offers automated statistical forecasting that selects the right forecast method that best forecasts the data.  It does this for each time-series in the data set.  This blog will help a laymen understand how the forecast methods are chosen automatically.

Smart makes many methods available including single and double exponential smoothing, linear and simple moving average, and Winters models.  Each model is designed to capture a different sort of pattern.  The criteria to automatically choose one statistical method out of a set of choices is based on which method came closest to correctly predicting held-out history.

Earlier demand history is passed to each method and the result is compared to actuals to find the one that came closest overall.  That “winning” automatically chosen method is then fed all the history for that item to produce the forecast.

The overall nature of the demand pattern for the item is captured by holding out different portions of the history so that an occasional outlier does not unduly influence the choice of method.  You can visualize it using the below diagram where each row represents a 3-period forecast in held out history, based on different amounts of the red earlier history.  The variances of each pass are averaged together to determine the method’s overall ranking against all other methods.

Automatic Forecasting and Statistical Forecasting App

For most time series, this process can accurately capture trends, seasonality, and average volume accurately. But sometimes a chosen method comes mathematically closest to predicting the held-out history but doesn’t project it forward in a way that makes sense.

Users can correct this by using the system’s exception reports and filtering features to identify items that merit review.  They can then configure the automatic forecast methods that they wish to be considered for that item.

 

 

Extend Microsoft 365 F&SC and AX with Smart IP&O

Microsoft Dynamics 365 F&SC and AX can manage replenishment by suggesting what to order and when via reorder point-based inventory policies.  A challenge that customers face is that efforts to maintain these levels are very detailed oriented and that the ERP system requires that the user manually specify these reorder points and/or forecasts.  As an alternative, many organizations end up generating inventory policies by hand using Excel spreadsheets or using other ad hoc approaches.

These methods are time-consuming and both likely result in some level of inaccuracy.  As a result, the organization will end up with excess inventory, unnecessary shortages, and a general mistrust of their software systems. In this article, we will review the inventory ordering functionality in AX / D365 F&SC, explain its limitations, and summarize how Smart Inventory Planning & Optimization can help improve a company’s cash position.   This is accomplished by reduced inventory, minimized and controlled stockouts.   Use of Smart Software delivers predictive functionality that is missing in Dynamics 365.

Microsoft Dynamics 365 F&SC and AX Replenishment Policies

In the inventory management module of AX and F&SC, users can manually enter planning parameters for every stock item. These parameters include reorder points, safety stock lead times, safety stock quantities, reorder cycles, and order modifiers such as supplier imposed minimum and maximum order quantities and order multiples. Once entered, the ERP system will reconcile incoming supply, current on hand, outgoing demand, and the user defined forecasts and stocking policies to net out the supply plan or order schedule (i.e., what to order and when).

There are 4 replenishment policy choices in F&SC and AX:  Fixed Reorder Quantity, Maximum Quantity, Lot-For-Lot and Customer Order Driven.

  • Fixed Reorder Quantity and Max are reorder point-based replenishment methods. Both suggest orders when on hand inventory hits the reorder point. With fixed ROQ, the order size is specified and will not vary until changed. With Max, order sizes will vary based on stock position at time of order with orders being placed up to the Max.
  • Lot-for Lot is a forecasted based replenishment method that pools total demand forecasted over a user defined time frame (the “lot accumulation period”) and generates an order suggestion totaling the forecasted quantity. So, if your total forecasted demand is 100 units per month and the lot accumulation period is 3 months, then your order suggestion would equal 300 units.
  • Order Driven is a make to order based replenishment method. It doesn’t utilize reorder points or forecasts. Think of it as a “sell one, buy one” logic that only places orders after demand is entered.

 

Limitations

Every one of F&SC / AX replenishment settings must be entered manually or imported through custom uploads created by customers.  There simply isn’t any way for users to natively generate any inputs (especially not optimal ones). The lack of credible functionality for unit level forecasting and inventory optimization within the ERP system is why so many AX and F&SC users are forced to rely on spreadsheets for planning and then manually set the parameters the ERP needs.  In reality, most planners end up manually set demand forecasts and reordering.

And when they can use spread sheets, they often rely on wide rule of thumb methods that results in using simplified statistical models.  Once calculated in the spread sheet these must be loaded into F&SC/AX.  They are often either loaded via cumbersome file imports or manually entered.   Because of the time and effort, it takes to build these, companies do not frequently update these numbers.

Once these are set in place, organizations tend to employ a reactive approach to changes.  The only time a buyer/planner reviews inventory policy is annually or at the time of purchases or manufacturing.   Some firms will also react after encountering problems with inventory levels being short (or too high).  Managing this in AX and F&AS requires manual interrogation to review history, calculate forecasts, assess buffer positions, and to recalibrate.

Microsoft recognizes these constraints in their core ERPs and understands the significant challenges to customers.  In response Microsoft has positioned forecasting under their AI Azure stack.  This method is outside of the core ERPs.  It is offered as a tool set for Data Scientists to use in defining custom complex statistics and calculations as a company wishes.  This is in addition to some basic simple calculations as a starting point are currently in their start up phases of development.  While this may hold long term gains, currently this method means customers start from near scratch and define what Microsoft currently called ‘experiments’ to gauge demand planning.

The bottom line is that customers face large challenges in getting the Dynamics stack itself to help solve these problems.  The result is for CFOs to have less cash available for what they need and for Sales Execs to have sales opportunities unfilled and a potential loss of sales because the firm can’t ship the goods the customer wants.

 

Get Smarter

Wouldn’t it be better to simply leverage a best of breed add-on for demand planning; and a best of breed inventory optimization solution to manage and balance costs and fulfilment levels?  Wouldn’t it be better to be able to do this on a daily or weekly basis to make your decisions closest to the need, preserving cash while meeting sales demand?

Imagine having a bidirectional integration with AX and F&AS so this all operates easily and quickly.   One where:

  • you could automatically recalibrate policies in frequent planning cycles using field proven, cutting-edge statistical models,
  • you would be able to calculate demand forecasts that account for seasonality, trend, and cyclical patterns,
  • You would automatically leverage optimization methods that prescribe the most profitable stocking policies and service levels that consider the real costs of carrying inventory and stock outages, giving you a full economic picture,
  • You could free up cash for use within the company and manage your inventory levels to improve order fulfillment at the same time as you free this cash.
  • you would have safety stocks and inventory levels that would account for demand and supply variability, business conditions, and priorities,
  • you’d be able to target specific service levels by groups of products, customers, warehouses, or any other dimension you selected,
  • you increase overall company profit and balance sheet health.

 

Extend Microsoft 365 F&SC and AX with Smart IP&O

To see a recording of the Microsoft Dynamics Communities Webinar showcasing Smart IP&O, register here:

https://smartcorp.com/inventory-planning-with-microsoft-365-fsc-and-ax/

 

 

 

 

Extend Epicor Prophet 21 with Smart IP&O’s Forecasting & Dynamic Reorder Point Planning

In this article, we will review the inventory ordering functionality in Epicor P21, explain its limitations, and summarize how Smart Inventory Planning & Optimization (Smart IP&O) can help reduce inventory, minimize stock-outs and restore your organization’s trust in your ERP by providing robust predictive analytics, consensus-based forecasting, and what-if scenario planning.

Replenishment Planning Features within Epicor Prophet 21
Epicor P21 can manage replenishment by suggesting what to order and when via reorder point-based or forecast-driven inventory policies.  Users may compute these policies externally or generate them dynamically within P21.  Once the policies and forecasts have been specified, P21’s Purchase Order Requirements Generator (PORG) will create automated order suggestions of what to replenish and when by reconciling incoming supply, current on hand, outgoing demand, stocking policies, and demand forecasts.

Epicor P21 has 4 Replenishment Methods
In the item maintenance screen of Epicor P21, users can choose from one of four replenishment methods for each stock item.

  1. Min/Max
  2. Order Point/Order Quantity
  3. EOQ
  4. Up To

There are additional settings and configurations for determining lead times and accounting for order modifiers such as supplier-imposed minimum and maximum order quantities.  Min/Max and Order Point/Order Quantity are considered “static” policies.  EOQ and Up To are considered “dynamic” policies and computed within P21.

Min/Max
The reorder point is equal to the Min.  Whenever on hand inventory drops below the Min (reorder point) the PORG report will create an order suggestion up to the Max (for example, if on hand after the breach is 20 units and the Max is 100 then the order quantity will be 80).  Min/Max is considered a static policy and once entered into P21 will remain unchanged unless overridden by the user.  Users often run spreadsheets to compute the Min/Max values and update them from time to time.

Order Point/Order Quantity
This is the same as the Min/Max policy except instead of ordering up to the Max, an order will be suggested for a fixed quantity defined by the user (for example, always order 100 units when the order point is breached). OP/OQ is considered a static policy and will remain unchanged unless overridden by the user.  Users often run spreadsheets to compute OP/OQ values and update them from time to time.

EOQ
The EOQ policy is a reorder point-based method.  The reorder point is dynamically generated based on P21’s forecast of demand over lead time + demand over the review period + safety stock.  The order quantity is based on an Economic Order Quantity calculation that considers holding costs and ordering costs and attempts to recommend an order size that minimizes total cost.  When on hand inventory breaches the reorder point, the PORG report will kick out an order equal to the computed EOQ.

Up To
The Up To method is another dynamic policy that relies on a reorder point.  It is computed the same way as the EOQ method using P21’s forecasted demand over the lead time + demand over review period + safety stock.   The order quantity suggestion is based on whatever is needed to replenish stock back “up to” the reorder point.  This tends to equate to an order quantity that is consistent with the lead time demand because as demand drives stock below the reorder point, orders will be suggested “up to” the reorder point.

Epicor Prophet 21 with Forecasting Inventory Planning P21

P21’s Item Maintenance Screen where users can specify the desired inventory policy and configure other settings such as safety stock and order modifiers.

Limitations

Forecast Methods
There are two forecast modes in P21:  Basic and Advanced.  Each use a series of averaging methods and require manual configurations and user determined classification rules to generate a demand forecast.  Neither mode is designed with an out-of-the-box expert system that automatically generates forecasts that account for underlying patterns such as trend or seasonality.  Lots of configuration is required that tends to inhibit user adoption and modification of the assumed forecasting rules defined in the initial implementation that may no longer be relevant.  There isn’t a way to easily compare the forecast accuracy of different configurations.  For example, is it better to use 24 months of history or 18 months?  Is it more accurate to assume a trend should be applied when an item grows by 2% per month or should it be 10%?  Is it better to assume the item is seasonal if 80% or more of it’s demand occurs in 6 months of the year or  4 months of the year? As a result, it is common for classification rules to be too broad or specific resulting in problems such as application of an incorrect forecasting model, using too much or too little history, or over/understating the trend and seasonality.   To learn more about how this works, check out this blog post (coming soon)

Forecast Management & Consensus Planning
P21 lacks forecast management features that enable organizations to plan at multiple hierarchy levels such as product family, region, or by customer.  Forecasts must be created at the lowest level of granularity (product by location) where demand is often too intermittent to get a good forecast.  There isn’t a way to share forecasts, collaborate, review, or create forecasts at aggregate levels, and agree on the consensus plan. It is difficult to incorporate business knowledge, assess forecasts at higher levels of aggregation, and track whether overrides are improving or hurting forecast accuracy. This makes forecasting too one-dimensional and dependent on the initial math configurations.  

Intermittent Demand
Many P21 customers rely on static methods (Min/Max and OP/OQ) because of the prevalence of intermittent demand.  Otherwise known as “lumpy”, intermittent demand is characterized by sporadic sales, large spikes in demand, and many periods with no demand at all. When demand is intermittent, traditional forecasting and safety stock methods just don’t work.  Since distributors don’t have the luxury of stocking only high movers with consistent demand, they need specialized solutions that are engineered to effectively plan intermittently demanded items. 80% or more of a distributor’s parts will have intermittent demand.  The stocking policies that are generated using traditional methods such as those available in P21 and other planning applications will result in incorrect estimates of what to stock to achieve the targeted service level.  As illustrated in the graph below, it isn’t possible to consistently forecast the spikes.  You are stuck with a forecast that is effectively an average of the prior periods.

Epicor Prophet 21 with Forecasting Inventory Management

Forecasts of intermittent demand can’t predict the spikes and require safety stock buffers to protect against stockouts.

 

Second, P21’s safety stock methods allow you to set a target service level but the underlying logic mistakenly assumes that the demand is normally distributed.  With intermittent demand, the demand isn’t “normal” and therefore the estimate of safety stock will be wrong.   Here is what wrong means: when setting a service level of, for example 98%, the expectation is that 98% of the time the stock on hand will fill 100% of what the customer needs from the shelf.  Using a normal distribution to compute safety stocks will result in large deviations between the targeted service level and actual service level achieved.  It is not uncommon to see situations where the actual service level misses the target by 10% or more (i.e., targeted 95% but only achieved 85%).

 

Epicor Prophet 21 with Forecasting Inventory Analytics

In this figure you can see the demand history of an intermittently demanded part and two distributions based on this demand history. The first distribution was generated using the same “normal distribution: logic employed by P21. The second is a simulated distribution based on Smart Software’s probabilistic forecasting. The “normal” P21 distribution recommends that 46 units is needed to achieve the 99% service level but when compared to actuals far more inventory was needed. Smart accurately predicted that 63 units was required to achieve the service level.

This blog explains how you can test your system’s service level accuracy.

Reliance on Spreadsheets & Reactive Planning
P21 customers tell us that they rely heavily on the use of spreadsheets to manage stocking policies and forecasting.  Spreadsheets aren’t purpose-built for forecasting and inventory optimization. Users will often bake in user-defined rule of thumb methods that often do more harm than good.  Once calculated, users must input the information back into P21 via manual file imports or even manual entry.  The time consuming nature of the process leads companies to infrequently compute their inventory policies – Many months and on occasion years go by in between mass updates leading to a “set it and forget it” reactive approach, where the only time a buyer/planner reviews inventory policy is at the time of order.  When policies are reviewed after the order point is already breached, it is too late.  When the order point is deemed too high, manual interrogation is required to review history, calculate forecasts, assess buffer positions, and to recalibrate.  The sheer volume of orders means that buyers will just release orders rather than take the painstaking time to review everything, leading to significant excess stock.  If the reorder point is too low, it’s already too late.  An expedite is now required driving up costs and even then, you’ll still lose sales if the customer goes elsewhere.

Limited What If Planning
Since features for modifying reorder points and order quantities are baked into P21 it is not possible to make wholesale changes across groups of items and assess predicted outcomes before deciding to commit.  This forces users to adopt a “wait and see” process when it comes to modifying parameters. Planners will make a change and then monitor actuals until they are confident the change improved things.  Managing this at scale—many planners are dealing with tens of thousands of items—is extremely time consuming and the end result is infrequent recalibration of inventory policy. This also contributes to reactive planning whereby planners will only review settings after a problem has occurred.

Epicor is Smarter
Epicor has partnered with Smart Software and offers Smart IP&O as a cross platform add-on to Prophet 21 complete with a bidirectional API-based integration.  This enables Epicor customers to leverage built-for-purpose best of breed forecasting and inventory optimization applications.  With Epicor Smart IP&O you can generate forecasts that capture trend and seasonality without having to first apply manual configurations.  You will be able to automatically recalibrate policies every planning cycle using field proven, cutting-edge statistical and probabilistic models that were engineered to accurately plan for intermittent demand.   Safety stocks will accurately account for demand and supply variability, business conditions, and priorities.  You can leverage service level driven planning so you have just enough stock or turn on optimization methods that prescribe the most profitable stocking policies and service levels that consider the real cost of carrying inventory. You can build consensus demand forecasts that blend business knowledge with statistics, better assess customer and sales forecasts, and confidently upload forecasts and stocking policies to Epicor with a few mouse-clicks.

Smart IP&O customers routinely realize 7 figure annual returns from reduced expedites, increased sales, and less excess stock, all the while gaining a competitive edge by differentiating themselves on improved customer service. To see a recorded webinar hosted by the Epicor Users Group that profiles Smart’s Demand Planning and Inventory Optimization platform, please register here: https://smartcorp.com/epicor-smart-inventory-planning-optimization/