Page Comparison

Goals/Scope

The reporting and analytics vision is to support OpenLMIS users with both routine reporting and ad-hoc analysis. Users need a way to extract and visualize data. This specification is not defining the exact list of reports needed but rather what types of reports (map, offline, etc.) and functionality is desired. Please see the Scope Out section for a clear explanation of what pieces this spec does not include like alerts (via email/sms) and integrations to systems like DHIS2.

This wiki page is an attempt to define the reporting and analysis needs for OpenLMIS users based on the following generalized user personas identified by the community.  

Image Added

Background

In an attempt to gain buy-in and agreement by the OpenLMIS community on what the needs, requirements and definitions are for reporting and analytics within OpenLMIS.  Do not focus on the technical solution as there may be multiple approaches to meeting the desired behaviors. Focus of this spec is to outline what the end users will need and want.  This spec will support the decision making process for implementing reporting and analytics within OpenLMIS.  

To support this conversation, we define reporting, dashboards and analytics in the following ways:

Reports, routine reporting and sometimes called built-in reporting contain detailed data in a tabular format and typically display numbers and text only, but they can use visualizations to highlight key data. Some key characteristics of a report include :

• It presents numbers and text in a table.
• It can contain visualizations (like a map, chart, graph) but only used to highlight findings in the data.
• It is optimized for printing and exporting to a digital document format such as CSV, Word or PDF.
• It is geared towards people who prefer to read data, for example, lawyers, who would rather read text over interpreting visualizations, and accountants, who are comfortable working with raw numbers.
• Rendered from a defined query/template.
• Does not require high technical capacity to run or access the report
• Usually viewed or run on a routine bases (more than once a year)
• Usually drives users to an action

Assumptions:

Designer is an engineer.

Routine reporting applies typically to more than one implementation.

Dashboards are a data visualization tool that displays the current status of metrics and key performance indicators (KPIs) for OpenLMIS end users. Dashboards consolidate and arrange numbers, metrics and sometimes performance scorecards on a single screen. Some key characteristics of a dashboard:

• All the visualizations fit on a single computer screen — scrolling to see more violates the definition of a dashboard.
• It shows the most important performance indicators / performance measures to be monitored.
• Interactivity such as filtering and drill-down can be used in a dashboard; however, those types of actions should not be required to see which performance indicators are under performing.
• It is not designed exclusively for high-level decision makers or but rather should be used by the general workforce as effective dashboards are easy to understand and use.
• The displayed data automatically updated without any assistance from the user. The frequency of the update will vary by organization and by purpose. The most effective dashboards have data updated at least on a daily basis.
• Does not require high technical capacity to access

Assumptions:

Designer is an engineer.

A dashboard applies typically to more than one implementation.

Analytics and Ad-hoc analysis leverages tools that offer the ability to select ad-hoc date ranges, pick different products, or drill down to more detailed data to be dashboards and my dynamic than reports. Some key characteristics of data analytics include : 

• user can define desired data element combinations (via a programming language or wizard).
• Not routine and the combination of data elements is run infrequently (once year).
• It fits on one screen, but there may be scroll bars for tables with too many rows or charts with too many data points.
• It is highly interactive and usually provides functionality like filtering and drill downs.
• It is primarily used to find correlations, trends, outliers (anomalies), patterns, and business conditions in data.
• The data used in a ad-hoc analysis is generally historical data. However, there are some cases where real-time data is analyzed.
• An outcome of ad-hoc analysis might be to define/identify performance indicators for use in reporting and/or dashboards.
• It is typically relied on by technically savvy users like data analysts and researchers with knowledge of and experience with report programming languages.

Assumptions:

Designer needs to be technically inclined to their BI&A tool of choice (e.g. Excel, DHIS2, Tableau, etc).  Per-seat licensing for report designer and consumer may apply depending on tool choice.

Typically more implementation specific, less re-use across implementations.

Above definitions were taken and adapted from a DashboardInsights article.

User Personas 

**If the implementer is using stock management, frequency of reports would increase.

Assumptions

• If the implementer is using stock management, frequency of reports would increase with transaction data. 

Project Status - Phase 1

Phase 1 Scope: DISC Indicators 1 and 2 (Full Stock Availability and Closed Vial Wastage) - Last Update

1/

16

•  

   AMI of base infrastructure setup and/or deployment scripts created

  •  

     Write Ansible scripts to install and configure a running instance

  •  

     Set up NiFi for data ingestion 

  •  

     Set up Kafka for real-time streaming & joining

  •  

     Set up s3 (HDFS) for historical indexing

  •  

     Set up Druid for storage and aggregations

  •  

     Set up Superset for visualizations

•  

   Load mock data through NiFi, index in s3, and into Druid

  • In progress -

    Village Reach team enriching

    Ona team will generate bulk demo data

    with Commodity Type and Trade Item data. All other data is enriched and loaded.
   

  • via Mockaroo. Clay meeting with Chongsun on March 5 to review Mockaroo, and then Ona will use Mockaroo to generate records

•  

   Working prototype of end-to-end system (OpenLMIS to dashboard)

  •  

     NiFi pull CSV/JSON data from a static hosted resource

  •  

     NiFi pulls data directly from OpenLMIS from 2-3 API endpoints, flows through Kafka, s3, Druid, to Superset for visualization

    • Build complete. Requires testing

  •  

     Write query for DISC 1 - Full Stock Availability in Superset

    • In progress - queries drafted

      , will refine

  •  

     Write query for DISC 2 - Closed Vial Wastage in Superset

    • In progress - queries drafted

      , will refine

•  

   Source code with open source license for reporting tooling and specific reports created

•  

   Documentation (developer documentation in the wiki or ReadTheDocs) to allow OpenLMIS Community to learn how to develop additional indicators and reports subsequent to the engagement

•  

   Reusable components that the OpenLMIS Community can use to build additional reports subsequent to the engagement

• If the implementer is using stock management, frequency of reports would increase with transaction data. 

User Stories

• Clay Crosby (Unlicensed) at some point let's link up the Jira work to the wiki pages. I know we are still getting through definitions but I do want to make sure at some point that is being put on the wiki.

Prioritized list of DISC indicators 

Definition of indicators: DISC Indicators.PDF

1. Full Stock Availability (defined as absence of stockouts between re-suppply periods) 

2. Closed Vial Wastage 

3. Stocked According to Plan

4. Forecast Demand Ratio

5. On-Time and In-Full Deliveries 

6. Functional Status of Cold Chain Equipment (Vidya Sampath, moved this down since we will have a screen to see which CCE is functioning and non-functioning) 

7. Temperature Alarm Rates  

Diagrams

Dependencies

Description

Link

Create user personas and outline objectives

Research of technical approach/options

Open Questions

Below is a list of questions to be addressed as a result of this requirements document:

#

Question

Outcome

Status

1

How much money does a MOH want to spend on resources to support creating report templates? We want to figure out guardrails on what types of budgets MOHs have to support the creation of reports. This will help us assess reporting tool options. For example, if one solution only allows for highly-specialized resources which cost a lot, we may not go with it. FYI, Josh Zamor.

In progress.

2

What type of skill sets can we expect implementers to have to create the report templates for routine reporting?

In progress.

3

Do users at the facility level want to be able to define their own reports? If so, what level of complexity would they be able to work with. Alfred Mchau could share some of his learnings as he digs into what the University of Dar Es Salaam did within eLMIS.

4

What is the approach if we have all the data within our systems?

What is the approach for when there is an aggregate HMIS (DHIS2) in place which we pull indicators from for resupply?

Example reports

Community members are encouraged to share key reports and examples for the team to review and keep in mind as moving forward.  Assuming the perspective of a national-level program manager who wants visibility into the activity of their program by region/district, but who would also like the option to get granular-level visibility into facility-by-facility data as well. 

Example reports made for ESMS

OpenLMIS ESMS Report examples.docx

1 - Stockout days 

• Ideally would be able to view on a map geographically 

• Could show avg. # of stockout days in a district – aggregated data from all facilities in that district to give a picture of the overall stockout rate for that district 

• User could click on the facility list and see the specific facilities reporting stockouts 

2- Stock levels/Consumption (these are two different indicators)

• Nice to view on a map geographically

• Shows SOH at beginning of period and SOH at the end of period for each facility 

• Graph could show min/max levels to indicate that stock delivery is adequate to bring stock levels up to max, if they’re ordering too much, or too little 

• Can show that a facility stock level is below the min or near stocking out 

• Should show value of commodity, even if the commodity is provided to the consumer for free, in order to reinforce idea that stocks have real value

3 -Timeliness and completeness of reporting 

• Could show aggregate for the district of whether district is reporting on time

• Similar to Stockout Report, user could click on the facility list and see the specific facilities which were not reporting on time 

4 - On Time and full deliveries 

• Could show aggregate data for a district/region for on time and complete deliveries that month. 

• Similar to Stockout Report, user could click on the facility list and see the specific facilities which reported delayed or incomplete deliveries

5 - Inventory aging

• Show items near to expiry (configurable date range, since length of the pipeline will determine how critical a given range is)

• Show items with VVM stage 2 status

6 - Expiry/wastage/Loss (based on reason codes)

• Show items that expired/damaged/lost and require disposal (units and value)

• Show loss rate; total lost / total usable stock used during period as percent (units and value) 

7 - Stocked according to plan (traffic light or similar indicator)--this is similar to but distinct from #2

• Show stocks that are within planned parameters (min/max or EOP/max, with config tolerance thresholds) for all commodities managed

• Show stocks that are below threshold for min/EOP

• Show stocks that are above threshold for max

8 - Forecast accuracy

• Show the ratio of actual consumption by product during a particular period compared to the consumption forecasted for the same period (central level dashboard)

9 - Emergency order rate

• Show number of unplanned orders/deliveries in period compared to planned deliveries

10 - Performance Score Card

• Compare HF/district/region against all peers for period and over time with a balanced scorecard comprising critical KPIs (configurable)

• This is both for self-assessment (How are we doing?) and for supervision/oversight (how are they doing?)

Some other ideas we tossed around but didn’t discuss as thoroughly were:

Financial Reports 

A program manager would use this report to measure how much money a facility should have (relating to cost recovery) 

A program manager could also see a graph depicting the total costs of requisitions for their area (get a clearer picture of where their money is going) 

Could see whether facility is in the red/black based on cost recovery analysis 

Should show value of inventory, value of good sold/issued/dispensed, value of goods lost

Emergency Orders 

Good to show on a map - district level making emergency orders to provincial level, potential risk of stockouts. 

Out of Scope

Data Model

Stock cards - each record / "row" in the database is a stock card line item, and will have the following data elements / fields / "columns" ("row" and "column" are used to show analogous terms to relational databases, whose structure is more familiar, although in this particular database we store the data as .json records)

• Stock card line item id ("id")
• Stock card id
• Occurred date
• Orderable name
• Orderable id
• Orderable code
• Commodity type
• Trade item
• Lot #
• Program name
• Program id
• Program status
• Facility supported programs
• Quantity
• Stock on hand
• Reason type
• Reason category
• Reason name
• Source
• Facility id
• Facility name
• Facility type
• District
• Region
• Region code - ISO code
• Country
• Ideal stock amount - value is added for all orderables within the corresponding commodity type + facility pair

Prioritized List of DISC Indicators & Business Logic for Each

Additional detail can be found in the initial design document: DISC Indicators.PDF

Implementation Considerations

The following items will need to be reviewed and potentially customized for your particular implementation's configuration

1. Product model hierarchy
   • Based upon the Commodity Types your instance uses, you may need to ensure the data model includes the proper tiers for your instance.
2. Geo Zone hierarchy
   • By default, we used the following Geo Zone hierarchy, but you may need to adjust this based upon the hierarchy in place at your instance:
     • Facility
     • District
     • Province / region
     • Country
3. Closed Vial Wastage
   • You should review the definition of line items that count as "closed vial wastage" ("numerator" column for Closed Vial Wastage in business logic table above) to ensure this definition comprehensively applies at your instance.
   • You should review the definition of "receipts" ("denominator" column for Closed Vial Wastage in business logic table above) to ensure this definition comprehensively applies at your instance.
4. Consumption
   • You should review the definition of "consumption" ("numerator' column for Forecasted Demand Ratio in business logic table above) to ensure this definition comprehensively applies at your instance.

User Stories

Diagrams

Dependencies

Example reports

Community members are encouraged to share key reports and examples for the team to review and keep in mind as moving forward.  Assuming the perspective of a national-level program manager who wants visibility into the activity of their program by region/district, but who would also like the option to get granular-level visibility into facility-by-facility data as well. 

Example reports made for ESMS

OpenLMIS ESMS Report examples.docx

1 - Stockout days 

• Ideally would be able to view on a map geographically 

• Could show avg. # of stockout days in a district – aggregated data from all facilities in that district to give a picture of the overall stockout rate for that district 

• User could click on the facility list and see the specific facilities reporting stockouts 

2- Stock levels/Consumption (these are two different indicators)

• Nice to view on a map geographically

• Shows SOH at beginning of period and SOH at the end of period for each facility 

• Graph could show min/max levels to indicate that stock delivery is adequate to bring stock levels up to max, if they’re ordering too much, or too little 

• Can show that a facility stock level is below the min or near stocking out 

• Should show value of commodity, even if the commodity is provided to the consumer for free, in order to reinforce idea that stocks have real value

3 -Timeliness and completeness of reporting 

• Could show aggregate for the district of whether district is reporting on time

• Similar to Stockout Report, user could click on the facility list and see the specific facilities which were not reporting on time 

4 - On Time and full deliveries 

• Could show aggregate data for a district/region for on time and complete deliveries that month. 

• Similar to Stockout Report, user could click on the facility list and see the specific facilities which reported delayed or incomplete deliveries

5 - Inventory aging

• Show items near to expiry (configurable date range, since length of the pipeline will determine how critical a given range is)

• Show items with VVM stage 2 status

6 - Expiry/wastage/Loss (based on reason codes)

• Show items that expired/damaged/lost and require disposal (units and value)

• Show loss rate; total lost / total usable stock used during period as percent (units and value) 

7 - Stocked according to plan (traffic light or similar indicator)--this is similar to but distinct from #2

• Show stocks that are within planned parameters (min/max or EOP/max, with config tolerance thresholds) for all commodities managed

• Show stocks that are below threshold for min/EOP

• Show stocks that are above threshold for max

8 - Forecast accuracy

• Show the ratio of actual consumption by product during a particular period compared to the consumption forecasted for the same period (central level dashboard)

9 - Emergency order rate

• Show number of unplanned orders/deliveries in period compared to planned deliveries

10 - Performance Score Card

• Compare HF/district/region against all peers for period and over time with a balanced scorecard comprising critical KPIs (configurable)

• This is both for self-assessment (How are we doing?) and for supervision/oversight (how are they doing?)

Some other ideas we tossed around but didn’t discuss as thoroughly were:

Financial Reports 

A program manager would use this report to measure how much money a facility should have (relating to cost recovery) 

A program manager could also see a graph depicting the total costs of requisitions for their area (get a clearer picture of where their money is going) 

Could see whether facility is in the red/black based on cost recovery analysis 

Should show value of inventory, value of good sold/issued/dispensed, value of goods lost

Emergency Orders 

Good to show on a map - district level making emergency orders to provincial level, potential risk of stockouts. 

Out of Scope

• Please note this spec does not include in application alerts (like an alert to an approver that a requisitions is ready for approval) which are addressed within each functional area (like Requisitions) based on business logic.  

• This spec does not include the scope of exporting data into another system like DHIS2.  If an implementation of OpenLMIS decides to use another application for all reporting needs, then the following feature set would not be used by that implementer.

Open Questions / Assumptions

1. Requirements document

   Below is a list of questions to be addressed as a result of this requirements document:

   #	Question	Outcome	Status
   1	How much money does a MOH want to spend on resources to support creating report templates? We want to figure out guardrails on what types of budgets MOHs have to support the creation of reports. This will help us assess reporting tool options. For example, if one solution only allows for highly-specialized resources which cost a lot, we may not go with it. FYI, Josh Zamor.		In progress.
   2	What type of skill sets can we expect implementers to have to create the report templates for routine reporting?		In progress.
   3	Do users at the facility level want to be able to define their own reports? If so, what level of complexity would they be able to work with. Alfred Mchau could share some of his learnings as he digs into what the University of Dar Es Salaam did within eLMIS.
   4	What is the approach if we have all the data within our systems? What is the approach for when there is an aggregate HMIS (DHIS2) in place which we pull indicators from for resupply?

   
   

2. Closed Vial Wastage denominator - reporting on the stock on hand at the start of the reporting period.

Consider the following data set where each row is a stock card line item. Note "quantity" is null for the first instances of each orderable id because stock on hand is the field we're anchoring this portion of the metric off of (and so quantity is irrelevant for our purposes here):

For a reporting period that spans the month of January 2018 (1/1/2018 - 1/31/2018), the stock on hand at the start of the reporting period would be 40 for orderable 1 and 50 for orderable 2. However, Druid and Superset cannot find data for rows with dates that are BEFORE the reporting period start date (1/1/2018). So, the stock card line item for orderable 1 which represents the stock on hand at the start of the reporting period would be excluded from the query as it has a date of 12/31/2017. There has been no stock activity on orderable 1 in facility 123. Note this date could be any date before the reporting period, but for the sake of simplicity we're using the day just before the reporting period.

To ensure that the Closed Vial Wastage denominator is accurately capturing the stock on hand for orderables whose most recent activity predates the reporting period start date, we will write a script to infer additional rows in Druid. These inferred rows would be duplicates of the most recent "real" row, with quantity set to 0 and reasons set to NULL to avoid erroneously counting the inferred rows when summing the quantity column or by counting rows with specific reason types for other indicators. These rows would be inferred for orderables attached to active programs during the dates for which the programs are active. I.e. if a program is active from 1/1/2017 to 6/1/2018, we would start inferring rows on 1/2/2017 (because there would need to be an anchor "true" row that we would infer from) and we would stop inferring rows on 6/1/2018. Because we want to support a daily granularity with this metric, i.e. users should be able to adjust the start date to 1/2/2018, we would do this inference each day.

So, extending the example above, the table would look like this for dates 12/31/2017 - 1/3/2018, where the italicized/emboldened rows are inferred rows.

Druid would be able to pull the value of the stock on hand column for each orderable on the reporting period start date because each orderable has a row for 1/1/2018.