Water Mission is a non-profit organization dedicated to bringing safe water solutions where needed around the world. The IBM jStart team is helping Water Mission to analyze water consumption data from Water Mission’s treatment stations in Uganda and Tanzania. Bluemix Apache Spark service is used to combine and correlate data with socio-economic and weather data, look for behavioral patterns, predict water shortages, and suggest changes to the way the stations are operated in order to increase delivery of safe drinking water within the local communities.
2. Water Mission
• Non-profit organization dedicated to bringing safe
water solutionswhere needed around the world.
• Founded in 2001, based in Charleston, SC
• Most of the projects in Africa, South America and
Asia
• 650+ communitydevelopment and disaster projects
currently underway
• 2000+ safe water and sanitation solutions built
3. Water Treatment Stations (Kiosks)
• Living Water™ Treatment Systems
• Making clean/safe water from otherwise
contaminated water sources
• Installed and monitored by Water Mission
• Maintainedby local communities
• Goal: make the maintenance sustainable
4.
5. IBM and Water Mission
• Started to work together in 2015
• Multiple on-going joined initiatives
– Alerting system
– Operational Dashboard
– Data analysis
• Focus on sustainability
6. Data Analysis
• Multiple data sources
– Water consumption data (all transactions)
– Historical weather data (including rainy seasons)
– Socio-economic data (income, occupations, agricultural seasons,
households, etc.)
• Comparing several communities
• Trying to answer some tough questions
– Can we predict water shortages and water outages?
– Can we predict drought conditions?
– What can be done to increase usage of the safe water in the communities?
• Some of these questions can actually be answered!
7. Sample Study
• 3 communities in Uganda
(Kikondo, Busiro, Dei)
• Kikondo – 1+ year in
operation
• Used Bluemix Apache
Spark service with Jupyter
notebooks to analyze the
data and come up with
recommendations on
increasing efficiency of the
kiosk.
8. Conclusions
• Add affordable program for replacing lost cards
• Conduct experiments with inserting more cards
into the system
• Consider issuing multiple cards to each
household