Doing good with data is very important to Immuta. We look up to great organizations such as DataKindwho organize data scientist volunteers, provide structure to events, and use the power of data science to help nonprofits be more predictive and effective. They are able to take data collected by these organizations, and use it in unexpected yet powerful ways — make an impact with data. We at Immuta also strive to help use data for good, and think our software positions us to help nonprofit organizations in unique ways. Data science always starts with data, and in the case of nonprofits, there is certainly cases where sensitive data and data collaboration between volunteers over a single weekend hackathon can be severely challenging. We want to take what we do for a living, and put it to good. Through our data knowledge collaboration tools, data abstraction layer, and privacy controls, we can aid data science events be more productive, or in some cases, happen at all due to the ability to control privacy constraints and policies.
Specifically, Immuta is currently planning a joint Hackathon event, managed by expert in Information Science, Franziska Wittleder; the event will combine participants from Ourselves, DataKind, and University of Maryland, in support of America’s Warrior Partnership (AWP). AWP works to empower communities to empower veterans by providing a community based program offering a proactive approach to serving Veterans. AWP uses a four-step approach: Connect, Educate, Advocate, and Collaborate. This is done by working through organizations in the Veteran’s communities and providing services. For example, a Veteran is searched out or met at an event, their information is gathered, and based on their needs, services provided. In doing so, AWP and its partners collect a vast amount of data on the Veterans but also collect follow-up information on services provided and the success of those services.
We, Immuta, will act as the data layer, providing the necessary policy controls to enable the sensitive AWP data to be shared with the volunteers (only see non-attributable data) as well as provide the knowledge sharing environment to work collaboratively over the weekend hackathon event. DataKind DC has discussed helping Immuta in this endeavor by marketing through their channels to drive more volunteers; as has the University of Maryland to bring data science students into the mix. We hope that the software and brainpower provided through this joint effort will forward AWP’s vision and goals, in this, and future events. Some very interesting questions that could be answered are:
- Provide new and potentially better factors to drive the current status of the veteran (how urgent are they needing help). For example, right now the status is driven by the Veteran’s immediate needs, but if there were better predictors hidden in the information captured, a Veteran with no needs today, could be deemed a worse status based on factors discovered through historical trends.
- Follow-on actions could also be better predicted.
More precise “tickle” follow-up time increments.
- When a service is provided (for example, employment found), can we determine the probability of success?
- If probability of success is low, are there services that could be provided to increase the probability of success?
- How can the AWP organizations more effectively find the Veterans in their area? Could we predict locations to advertise for events? Doors to knock on?
- Transportation seems to be a key component to success in employment and housing services. Can we add/join spatial factors such as location and public transportation to determine probability of success?
- Similar to transportation, how can other features be joined to the currently collected data to enhance the models. For example, could we categorize the companies where Veterans are placed as a factor for success?
- How can Veterans be “tickled” for more out-of-cycle data, meaning, can meaningful features be captured from the Veterans even when they are in good status and do not have a current running case?
- Probability or predictor of Homelessness?
- Track Likelihood of attending events and what type of events.
We’ll keep you updated on the progress of this event through our blogs and website and please feel free to join in on the effort in any way you can. Reach out!