2 Demand Forecasting Dashboard
The IPOCS-Demand Forecasting Dashboard is the workspace that is seen upon login. You can use the dashboard to quickly analyze the health of the business. The dashboard will need to be refreshed periodically as new products, locations, and demand-related data are added. This typically happens weekly, but depends on your administrator's settings. The measure data of existing products and locations in the dashboard can be refreshed at any time to view updated changes.
Note:
For additional information about the user interface, dashboard, and workspaces, click Application Help from the Help icon on the dashboard.
IPOCS-Demand Forecasting has two dashboard profile types and includes two profiles for each type:
-
Metric profile type
-
Exception profile type
Figure 2-1 Dashboard Profiles
The information in the metric and exception profiles is summarized in tiles, which can be filtered on the product, location, or any other hierarchy that makes sense as shown in Figure 2-2 and Figure 2-3.
Figure 2-2 Product Filter
Figure 2-3 Location Filter
The information that is presented condensed in the tile metrics, is also presented in a chart with more detail.
In addition there is the Administration dashboard which gives information about scheduled OAT tasks, or workspace build information. More about the Administration dashboard can be found in the Oracle Retail Predictive Application Server Cloud Edition User Guide.
Overview Dashboard
There are five GA tiles in the Overview dashboard.
Figure 2-4 Overview Dashboard
Note:
This view has a dual Y-axis and the scales of the Y-axes could be different from each other.
Current and Historical In-stock Rates
The first tile displays information about how often stockouts occurred. If every four weeks, a stockout is registered, the in-stock rate would be 75%. This information is driven by how often stockouts occur. The tile displays current and historical fill rates.
The in-stock calculations are:
In-stock% = (total number of periods - periods with outage) / total number of periods number of periods for the Current version of the metric: given by Current Period Window in Forecast Setup/Exception Management step number of periods for Historic version of the metric: given by Historic Period Window in the Forecast Setup/Exception Management step.
Stockouts
The second tile displays information about the current and historical fill rates, which are related to service levels. This information is directly tied to the amount of lost sales caused by total or partial stockouts.
Breakdown of Sales
The third tile shows the breakdown of sales. The total sales are the sum of regular, promotional and markdown sales. The promotional and markdown sales are displayed as secondary metrics to gauge the amount of sales driven by these extraordinary events.
Promotional and Markdown Sales
The last two tiles display information about promotional and markdown sales. Besides the total units, additional information is displayed as secondary metrics. For instance, the tiles display how often the merchandise are promoted or marked down, as the promo or markdown frequency. Also, it shows the effectiveness of how promoting or marking down merchandise is increasing demand as compared to regular demand.
The calculations for the Markdown Effectiveness and Frequency are:
Markdown Effectiveness = (sum of markdown units) / (sum of regular units)
Markdown Frequency = (number of periods with markdown sales) / (total number of periods)
The number of periods used for these calculations is given by Recent Sales Window in Forecast Setup/Exception Management step.
It may seem a little confusing to introduce yet another time window, but if we are using just the current period window, all the periods may have markdown sales, which may not be a complete picture of the business. This is why markdown has its own controls.
Forecast Scorecard Dashboard
The information displayed in this dashboard helps the forecast analyst or forecast manager determine how accurate the forecasts are.
Figure 2-5 Forecast Scorecard Dashboard
Error Metrics
These are the error metrics:
-
Mean Absolute Percentage Error
-
Root Mean Squared Error
-
Mean Absolute Error
-
Forecast Bias
-
Percent Adjusted
The error metrics are detailed in Appendix: Forecast Errors in the Forecast Scorecard Dashboard. They are calculated for both the user-adjusted, as well as for the system-generated forecasts, allowing a fair evaluation of the performance of the analyst versus the system.
Forecast Bias
Another tile is dedicated to the forecast bias, a useful metric in determining if the forecast is consistently over or under the actual demand.
Percent Adjusted and Adjusted Volume
Finally, there is the Percent Adjusted tile, with the Adjusted Volume as the secondary metric. Here is displayed the percent of adjusted forecasts, and the volume resulted as the result of the adjustments. For instance, if the percent adjustment is high, but the adjusted volume is not significant, the analyst may want to spent time and effort on other areas of the merchandise.
Approval Alerts
This is the exception profile that shows the result of the approval business rules.
Figure 2-6 Approval Exceptions
Exceptions
The information displayed in the tiles is based on exceptions, or business rules defined to check if the forecasts are within bounds with respect to some thresholds. The exceptions definitions and expressions are detailed in Appendix: Forecast Approval Exceptions.
Number of Hits Count
The main metric in the tiles is the number of hits count which is based on the product and location filter settings. If the filter settings range products and locations to a narrower selection than, All Product and All Location, then the number should be reflected in the tile; that is, it should decrease.
Number of Subclass and Districts
The secondary metric on the tiles display the number of subclass and districts that are affected. Note that the dimensions are configurable.
Charts
Each tile comes with a set of charts where the hit count information is broken down to lower levels and sorted based on hit count, or average variance.
Note:
The average variance represents how much off the forecast is versus its target. For example we compare the forecast against last year sales. If the sales are 10 units and the forecast is 15 units, the variance is 5 units. The average variance is the average of all the differences between forecast and last year sales.
The hit count represents how many times the forecast did not pass the configured business rules. The average variance represents by how much the forecasts were off. This is useful, because if a subclass is heavily alerted, but the forecasts only barely miss the threshold, the user may want to concentrate on other merchandise, potentially less alerted, but where the forecasts were off by a larger amount.
Navigations Alerts
The approval exceptions are very helpful in deciding what product/locations are approved and which ones need review. However, reviewing unapproved product/locations in order of critically, makes the reviewing process much more efficient. You can start reviewing Urgent items, then move on to Required, and if time permits you can review Optional and then Informational.
Figure 2-7 Navigation Exceptions
Exceptions
The product/locations presented in this dashboard are virtually the same as the ones in the Approval Exceptions. The difference is that the product locations are sorted in order of priority, which is a proxy for the importance to the business. The rules to determine the priorities are set up in the Business Rule Engine, as detailed in the Appendix: Business Rule Engine.
Forecast Review Workspace
Finally, when a user makes a decision as to which merchandise and locations to review, the user can launch directly in the Forecast Review workspace by clicking Open in Workspace, with the locations and product ranged down to the desired selection. Once in the workspace, you can choose to use the navigation alerts to navigate to product/locations. Every navigation exception can have a workspace counterpart. Using these exceptions you are guided to the exact product, location, and time period that needs your attention.
The Open in Workspace button is located in the right corner above the top chart
Note:
Alternatively, you can right-click on any exception chart and select Open in Workspace