Flink Meetup #8 Data Flow vs. Procedural Programming & Interactive Data Analysis
Details
http://photos2.meetupstatic.com/photos/event/7/9/d/0/600_438691184.jpeg
Join us at the 8th Apache Flink (https://flink.apache.org/) Meetup,
drinks and sandwiches sponsored
by data Artisans (http://data-artisans.com/).
Talks
- Flink Community Update
By Márton Balassi
- Data flow vs. procedural programming: How to put your algorithms into Flink
By Mikio Braun
Modern Big Data frameworks including Flink are often based on a data flow programming model. The main data type is a set, and an algorithm must be formulated in terms of transformations on these sets, dealing with one element at a time. This is in stark contrast to classical programming languages, which are based on variables, functions, and
control flow like for loops and conditional statements to process data.
Mikio will discuss both approaches and show how to translate a more classical piece of code into the data flow formalism, to be able to benefit from the scalability of these systems.
- Interactive data analysis with Apache Flink
By Till Rohrmann
Data analysis becomes more and more important for companies and institutions alike as they gather and ever increasing amount of data. Extracting useful information from this data requires elaborate tools which have to scale well with the data. Furthermore, data scientists require explorative tools which allow them to run queries against their data on an ad-hoc basis and to visualize the corresponding results. This helps them to gain new insights more quickly and to communicate their findings more easily to other people.
Apache Flink offers a solution to these problems by providing a rich machine learning library which can be used from within Flink's new interactive shell. Moreover, Apache Zeppelin lately added Flink as a supported backend which brings an IPython Notebook-like interface to the Flink world.
Till will demonstrate in his talk how we can use Flink's machine learning library to solve a data analysis task at large scale. He will also show how we can use the interactive shell to explore our data and how it can be visualized using Apache Zeppelin.
---------------
Bring your data
After the talks, while having a drink, there's the opportunity to work together with Flink committers on an interesting data problem you're facing.
Please contact Kostas Tzoumas at kostas@data-artisans.com if you're interested in taking part in this!
---------------
About Till
Till is a committer and PMC member of Apache Flink.
His main work focuses on enhancing Flink's scalability as a distributed system and building a large-scale machine learning library with Flink. Till also contributed to Apache Mahout and helps presently to add Flink support to the Mahout DSL. Till earned his MS in computer science from Technische Universität Berlin where he focused on machine learning and massively parallel dataflow systems.
About Mikio
Mikio Braun is a data scientist and PostDoc researcher at TU Berlin. Mikio is interested in scalable and realtime machine learning and is also author of jblas, a fast linear algebra library for Java.
About Márton
Márton Balassi is a committer and PMC member for Apache Flink, currently working on developing Flink at data Artisans. His main focus is the stream processing component. Márton has presented Flink at ApacheCon, Hadoop Summit and at numerous meetups lately.
-----------------
SCHEDULE
19:00 - 19:30: Sandwiches and Drinks
19:30 - 19:45: Flink Community Update
Marton Balassi
19:45 - 20:30: Data flow vs. procedural programming
Mikio Braun
20:30 - 20:45: Break
20:45 - 21:30: Interactive data analysis with Apache Flink
Till Rohrmann
21:30 - end: Socializing and Drinks
http://photos1.meetupstatic.com/photos/event/a/1/2/0/600_438701248.jpeg
Flink Meetup #8 Data Flow vs. Procedural Programming & Interactive Data Analysis