Datathon for Social Good: IBM Z & UC Berkeley
A special thanks to the team : Dr. Alex Osadchyy, Elizabeth K. Joseph and Teja Karra who helped pull the event together and write this blog!Fri, Nov 8th and Sat, Nov 9th was our first ever Datathon event at UC Berkeley. “Datathon for Social Good” was an IBM sponsored event, hosted by UC Berkeley’s Data Science Club.
Our team had worked on this for months, meticulously planning activities to educate and engage students about IBM Z and ML on Z. I’m sure you can imagine how pleasantly surprised we were when we walked into the event on Friday, Nov 8th to – a room full of students – over 100 students (standing room only). Students were probably coming straight from class to find out what this “Datathon for Social Good” was all about.
After brief introductions, we got straight to business – our first act in the agenda was a Panel Discussion about IBM Z – the Modern Mainframe. Our panel was star studded with Damyon Thompson, Elizabeth K. Joseph, Teja Karra from IBM & Veronica Gilliard from California Dept of Technology, State of CA. The panel was moderated by Sudharsana Srinivasan of IBM. It was eye opening to see that almost no student had heard about IBM Z or Mainframe. When asked if they had used a mainframe – Silence in the room. When asked if they had used their credit or debit card for the day – a lot of nods and show of hands. That’s when the students had a moment of epiphany – they were Mainframe users – IBM Z users. The panel discussion went into details about how IBM Z infrastructure is different from the traditional x86 servers, how the hardware has security built in (from the Hardware Security Modules to cryptographic co-processors), and how it is scalable real time, vertically, with ease. Elizabeth, coming from the Open Source world with a lot of x86 experience, was able to articulate to the students how impressive the speed and reliability of IBM Z, and how it is called IBM Z for Zero downtime, because it’s enterprise-grade hardware that has a backup for everything, unlike the standard x86 blade servers she had worked with in the past. Veronica from State of California reiterated how critical mainframes are to our everyday lives. She shared with students that this is a highly sought-after skill and is actively looking for next generation of mainframers!!
Dr. Alex Osadchyy then took the stage to talk to the students about the theme of the Datathon – Machine Learning on IBM Z and Social Good problem areas such as Homelessness. Historically IBM has been leading in data analytics, ML and AI such as with IBM Watson. There are a number of products available on IBM Z such as IBM Open Data Analytics on z/OS (IzODA) and IBM Watson ML on z/OS. The main story for Data Scientist students was to position IBM Z as the best production platform for their models. They could design a model using IBM Watson Studio in IBM Cloud or on IBM Z. Mock hundreds of various algorithms, visualize them, test etc. Then when it's ready - deploy on IBM Z for production use. IBM Z brings benefits of data proximity, real-time insights (several ms vs tens-hundreds ms), availability, security and throughput. Dr. Alex Osadchyy provided a Demo of the IBM Z platform students would use for the challenge.
Saturday came along and we had 16 teams working away from 7:00 a.m to meet the submission deadline of 3 p.m. These students learn fast, that’s for sure. They learned that IBM Z is great to crunch loads of data fast, highly scalable and a very secure platform. Several teams leveraged IBM Z as the platform for their solution. Here is the list of winning projects :
1st Place - GaMHErs with student financial help solution
Financial planning application, Pocket, which utilizes IBM Z to update and analyze users' budget allocations in real-time based on spending patterns and provides relevant daily insights into federal and campus resources using machine-learning algorithms.
2nd Place - Fire with wildfire modelling.
Using SIR to predict where and when fires may spread which utilizes IBM Z to run the data to ensure speed and accuracy.
3rd Place - Cartwheel with Homelessness support spending.
Re-allocate budget based on main drivers to homeless improvement
This event would not have been possible without the help from our amazing LinuxONE Community Cloud guru Bruce Gilkes and team. This team did the magic only they know, to ensure we had a platform that could scale on the day of the event.
Thank you to the Data Science Club at UC Berkeley for putting this event together. It was a super well planned and smoothly executed event. With ample food and drinks!!
Watching students learn about IBM Z was so much fun…. We would love to do many more such events!