GSCTechCom16

Data & Analytics Unconference
Live from CSC: updates, insights & proceedings about tools, projects, writer's block & more
Davor Andric
Prediction Extension for Rail (PXR)
is a Smart Analytics implementation based on our IoT, Big Data & Cloud Capabilities / Products. PXR are based on current Project for Timetable prediction in Real-Time for the whole German rail network through Big Data.
Junjie Tang
is it a framework or product that we can easy transfer it for other counties?
Davor Andric
Yes, PXR have a Services that can be consumed by different Clients over Coutry boundary's.
Davor Andric
Main PXR Services are: Time Table Prediction Service, Common Delay / Error Detection Service
Faisal Siddiqi
How to do behavioral analysis of social media data (Karthik)
Faisal Siddiqi
First identify the social media data - for example Tweets, FB Pages, relationship graphs
Faisal Siddiqi
What can we analyze?
Faisal Siddiqi
Measure activity level
Faisal Siddiqi
Classifying by interest
Faisal Siddiqi
Trends analysis across users over time
Faisal Siddiqi
Python is a great platform for collecting and analyzing social media data
Faisal Siddiqi
@srini_varadhan MEAN stack to query twitter, weather APIs, flight information - smart trip planner on Azure
Pavel Hruby
Data entry - a blog about how professionals (healthcare, manufacturing) are not trained in it (Elise Veltman)
Pavel Hruby
Healthcare - often garbage in is garbage out - data entry is one of the reasons
Pavel Hruby
Chinese connection = swivel chair
Joost Platenburg
= draaistoel interface (Dutch)
Pavel Hruby
Elise started her session by asking all participants: what do you think this session is about, based on your understanding of the title. This sets the agenda - great start.
Pavel Hruby
We need to rethink the user interfaces - especially for data entry on mobile devices.
Pavel Hruby
We extend this session by 20 min - following sessions shift to the next slot. Excel sheet with program has been updated.
Joost Platenburg
Mobiel divices are not always the tablets or mobile phones, but can also (in healthcare) be things like beds, camera's or other sensors available in logterm care facilities.
Pavel Hruby
We need the Internet of Things point of view, and use is it to design new user interfaces.
Pavel Hruby
Do we need GUIs at all?
Pavel Hruby
How are people going to learn all that? Gamification?
Elise Veltman-van Reekum
Thank you Pavel for this great overview of a really interactive session! And thank all the participants for changing my mind :)
Srinath Komandur
CSC has been providing testing services for over a decade. A statistical defect prediction model would help to anticipate defects and plan/estimate work accordingly
Karthikeyan Ganesan
Innovation lab -
CSC managing 100 Million Patient. How to produce nice Analytical information using this Healthcare Data.
Jerry Overton
Would love to get access to this and take a crack at it
Junjie Tang
lovely. Can you talk about the security perspective?
Pavel Hruby
Data Democratization (Ravindra)
Pavel Hruby
Common data model for healthcare? Comments would be greatly appreciated.
Pavel Hruby
Can we write guidelines from experience we have in this area?
Joost Platenburg
The persons to talk to (in the Netherlands) are Bob Schat for the common datamodel for healthcare, he has co-written a book about this. If talking about consent models for sharing the medical data is Frank van Grinsven.
Jerry Overton
Case Study: Prediction in Rail, the PXR project, 12:15 pm- 12: 45 pm
Jerry Overton
Just got a volunteer to translate the customer story slides from German to English. Can't wait to see this posted to C3.
Jerry Overton
Algorithms were created by collaborating with PhDs from universities
Jerry Overton
The opportunity came through co-operation with the vertical. Interesting. Domain knowledge was key.
Hassan Nasser
so the Big Data platform as a service Platform is using our InfoChimps?
Bijoy Abraham
PXR project -Great example of Big Data, IOT, data science project. Congrats to the team
Davor Andric
Good point @JerryAOverton we are currently on it.
Davor Andric
@JerryAOverton we hire PhD in our BD&A DS group
Junjie Tang
How can you improve the performance of realtime prediction? Short training period is one way
Davor Andric
Can be (1) Paralelisation by Alogo. Optimisation ML (2) usage of GPU
Faisal Siddiqi
What tools can we recommend to SQL developers for using hadoop
Faisal Siddiqi
Start with Pig to procedurally do relational processing with HDFS data
Faisal Siddiqi
HIve for running SQL like queries against HDFS data
Mark Perry
DATA EXPLOITATION VERSUS DATA PRIVACY. Good discussion on the balance needed and methods which we capture in a white paper shared with stakeholdes for example: enterprise-wide data modelling with clear data ownership; understanding & monetising data value
Jerry Overton
Sounds quite interesting