nudgeAI

• 

  Shape of data and dataset sizes

• 

  Nudge scores predicted and error variance. we use predicted score to nudge user to behavioral actions

Inspiration

When learning a new skill, users engagement drops over time. Companies currently send push notifications to ask users come back, but this is ineffective.

What it does

NudgeAI uses science backed human behavioral cues to assist users to finish a learning journey they began.

How we built it

Using existing datasets that tracked user engagement over a long period of time, we used Python/Azure Functions/Azure App services to predict motivation score using commonly measured data metrics such as time on screen, click thru rates, notification responses etc., it’s a Nobel way of solving a well known problem.

Challenges we ran into

Gathering real world dataset ( took almost a day), marrying behavior psychology with tech metrics to nudge users.

Accomplishments that we're proud of

nudgeAI is made available as an API endpoint which can be used any elearning company (DuoLingo, LinkedIn, PluralSight, Masterclass, Goverment initiatives etc.,) to actively engage customer there by increasing customer engagement and help users to finish a course/task they started.

What we learned

Idea came from most unexpected places, I had random conversation with a neuroscientist who dropped by to check out what’s going at AISF. we learned solutions to common problems in tech world existing in adjacent verticals.

What's next for nudgeAI

Scale up servers, provide endpoints for eLearning companies to test/roll out scores. Expand nudgeAI capabilities to other verticals (fitness, productivity, sustainability, fintech (savings)) we need a lot of guidance for this.

Built With

• azure
• data-science-toolkit
• numpy
• pandas
• python
• scipy