Kebibji

Inspiration

We found that the data-set contains information about a lot of sources and a lot of sensors. It is hard to keep watching every-one of them, rather we thought of a system that list and visualize data from all the sensors that reported an outlier value whenever the system asked to do.

What it does

A mobile app that plot a time series plot for a sensor values versus the time. It also can mark where in the graph a sensor value is considered as outlier using z-score calculation.

How we built it

Mobile app with android studio.
Outlier detection with Python.
Treating the data csv files as our database and using pandas to manipulate, calculate the outliers. Then, using seaborn for visualization.

Challenges we ran into

Lot of ideas, and only 36 hours.

Accomplishments that we're proud of

We worked in the week-end :P.
Communication overseas was a challenge, yet we managed to use the gather.town and discord to communicate ideas, and code effectively.

What's next for Kebibji

The mobile app is more of prototype or proof of work. The app don't really connect to a database or a server to do the calculation, rather every component in the system is isolated for now and no communication between them.
- Notification system when a sensor value is detected as anomaly, or a sensor didn’t report its value for a specific period of time.
- So far the system detect the anomalies, you investigate. But, what if we have smarter system with more data about the lab where the microscopes are operating.