This abstract has been accepted at the 2024 DHIS2 Annual Conference
DHIS2 usage analytics: A data warehouse approach to understanding data use
How are the end users navigating the system? What are the main visualizations accessed? Are the dashboards really being used? All these are key questions in order to measure how the system is being used, and if the system is having an impact in terms of data quality, access and meaningful visualization. DHIS2 has tools (eg. Usage Analytics) that allow rendering visualizations for metrics such as favorite views, top favorites, etc. However, depending on the complexity of the metrics to render, the default DHIS2 apps are not currently as adequate at measuring dynamic analyses of different types of user behaviors and engagement. Most of this information can be exposed in the system logs as free text, especially in the web server and application logs. It is possible to find in the navigation of those files information about the most common API endpoints used, latency time, devices and browsers accessing the system, and, if exposed, user names. For complex usage analytics we would like to share an approach that uses the DHIS2 system logs as a source for analyzing end user behavior and engagement with the system. Regardless of the size of the log files and their format, it is possible to transform the content of those files in dimensions and facts that can be stored in a data warehouse. It is then possible to connect the data warehouse with any BI tool (like PowerBI, Tableau, superset, etc.) to perform complex calculations, generate custom metrics, and render advanced visualizations of DHIS2 system use. We would like to share what we have learned in terms of designing global data warehouses for usage analytics (eg. what are the most important common dimension tables and how they are related to the facts). Most importantly, we would like to demonstrate what kind of results and analytical outputs you can achieve analyzing the logs of the DHIS2 servers, showing several example from large-scale DHIS2 implementations. We will highlight considerations regarding security and data privacy (especially relevant when we need to analyze user actions and behaviors). We conclude by sharing what types of usage data we have found most useful across real implementations to inform future discussions about the DHIS2 roadmap for analyzing user engagement and data use.
Primary Author: Carlos Tejo Alonso
Keywords:
DHIS2, usage analytics, data warehouses
5 Likes
Hi!
Wow, this is something I know many community members have asked about! 