Forecasting and Estimation

Introduction

Annual national forecast needs to be entered into OpenLMIS rather than building in the various forecasting logic required, especially because there are different methodologies and sometimes mixed methods for determining forecast. But some information must also be included as text to capture the methodology used, the assumptions, and any constraints that affected the final quantity forecasted (e.g. budget limits or storage capacity). Forecast should be the true "constrained forecast" on which supply plans and forecast accuracy are based. OpenLMIS should enable upload of forecast for all levels if the program follows the traditional WHO methodology.

If the annual forecast is used for allocation/push resupply to lower levels (the traditional WHO methodology), this process can be used for either routine or campaigns, since the target-based methodology is the same. However, campaigns are otherwise out of scope for MVP

Specific forecasting and estimation features planned for OpenLMIS version 3.3

This table shows the specific features and user stories slated for inclusion in the OpenLMIS version 3.3, which will include the MVP for managing vaccines. The top row indicates the key user activities within the domain, click on the header to navigate to page with details on the related user stories (vertical cells in the table). The user stories with green backgrounds are currently planned for the MVP for the development roadmap. The stories with yellow backgrounds are under discussion and may or may not be included in the MVP scope. The stories with white backgrounds will be considered up next if more funds become available or after the initial release.

Legend

Green is planned for MVP
Yellow needs discussion
White is up next
Headers = Key user activities
Cells = Stories

Calculate/Enter Product Need/Estimate	View Forecasted Product Needs	Calculate Pipeline	View Pipeline
Provide upload file option for annual needs forecast (i.e. needs calculated externally)	View forecasted needs for each product and multiple time periods	Calculate central future stock position Down Vote: 1	View pipeline
Modify annual needs forecast/ targets as a DIVO Down Votes: 2	view forecasted needs each storage point by product and time period Votes: 8	Calculate intermediate level future stock position Votes: 3
Create rights for managing annual needs forecasts or Ideal Stock Levels	Reorder Calc in Requisition Votes: 4
WHO formula calculator Votes: 5
Aggregate forecasted product needs for campaigns Votes: 2
Display the formula used for calculation to the viewer
Provide a calculation demo to viewer to confirm accuracy of calculation
Enter forecasted product needs for campaigns → enter SDP microplan forecast for campaign into OpenLMIS campaign screen
Provision of extension points to countries to customize needs estimation on their own
Provide all levels an option to correct annual needs forecast (i.e instead of set annual population targets)

Source: Final MVP whiteboard from Vaccines Workshop: Vaccine Workshop Notes and Whiteboards - May 2-4 2017

Out of Scope

National forecast planning process

Open Questions

Question	Outcome

Detailed Scope Activity

Background

Ashra Islam (JSI), Gemma Orta-Martinez (UNICEF), Jorgen Kofoed (UNICEF), Alpha Omari (JSI) attended the breakout session at the OpenLMIS Vaccine Module Workshop.

Reporting Outcomes

KPIs:

Forecast Accuracy (National level is a must, lowers levels if resupply is by allocation/push)
Stock at Risk (projected stock outs or expiry alerts by location, antigen, and $ value) (all levels)

Reports:

Forecast per antigen based on any methodology: demographic, session size, historical consumption trend, or a mix of these (e.g. demographic first, adjusted by historic consumption trend to keep it reality based in order to avoid overstock).
1. If using consumption, must adjust for reporting rate from facilities if less than 100%
Forecast accuracy, by level, time period, antigen, volume/quantity, and $ value. Compares forecast to actual consumption (adjusted at national level for reporting rate) for a specific period of time, and presents the data as a percentage (e.g. 93% accuracy means actual consumption is 93% or forecasted consumption). Also displays the $ value of the difference (7% below forecast = $X.00 worth of product(s) not used)
Pipeline report (national level is a must, other levels nice to have if allocation/push system): SOH, future stock position based on AMC rate (including coverage and wastage), and planned inbound shipments.
1. AMC must take into account reporting rate from facilities if less than 100%

Dependencies

Census data (dis-aggregated to locality)
Birth cohort Data by SDP
Average session size data by SDP
Historic consumption data by SDP
Antigen Dose per vial
Immunization schedule
target coverage
reporting rate (% reporting)
open vial wastage rate
closed vial wastage rate
national stock status (whole pipeline)
forecast accuracy for adjustment
previous order fill rate (esp. from supplier UNICEF)
Funding availability