This talk will describe graph based data modelling and analytics as a means to help organizations figure out the various nuances and hidden elements within their current data models. It will also delve into the various techniques and approaches that will enable them to leverage these data systems. It will cover key questions that organizations typically face: Why should they move to Graph based data modelling? When do they need to start migrating to the Graph paradigm? And How to do this transformation to build analytics from simple aggregations to complex machine learning based analytics?
Large Scale Graph Based
What, Why, When, How
A Step by Step approach
• The talk will discuss about what do we actually mean when we say
“Graph Based Large Scale Data Analytics” and what we can achieve by
• We will discuss how data and its inherent relationships are first class
citizens in the world of graph based analytics
• What are different use cases of Graph Analytics??
• For example Aggregations, Pattern Mining, Graph based machine
• After having understood the “What”, we will delve little deeper in the
subject asking ourselves “why” we need it?
• We will tackle this by using a example walk through by describing the
problems with existing analytics systems(RDBMS for example!) and
how they can dealt with elegance, ease and flexibility using Graph
• We will try to understand in this section “why” it is critical to use both
data and relationships to extract maximum value out of current data
and its inherent relationships.
• Having understood the “what’s” and the “why’s” we will delve even
more deep into the subject by asking ourselves the question “when”?
• When do we bring in Graph based analytics in our analytics
• What are the hints our current paradigms are showing which can
point or hint towards migrating towards Graph based analytics
• We will take a walkthrough based approach to explain for looking for
such hints and pointers in the discussion…
• Having built the context by explaining “what”, “why” and “when”, the
most important step is “how”…
• How to bring in Graph based analytics system in existing analytics?
• First…How to bring about Data Modelling changes from existing Data
Models(for example RDBMS based) to Graph Based?
• How to do this migration efficiently and effectively?
• How to migrate existing aggregation and predictive analytics from one
existing data model to Graph based data model?
• This will be show cased with an example or walkthrough…
• Having built the use case for migration to Graph based analytics by
explaining “what”, “why”, “when”, “what” and successfully migrating
existing analytics to graph based systems…we have only scratched the
surface with what all can now be achieved by using this paradigm.
• We will go further deep into the subject
• What are the design principles and patterns for doing right kind of
Graph based Data Modelling..
• What are different Graph algorithms that one should be aware of and
• Algorithms: Path Analytics, Connectivity Analytics, Community
Analytics and Local Properties, Global Property: Modularity, Centrality
• Graph Analytics Applications
• Hands on execution of above algorithms and using Neo4J Cypher or
Finally walking in the distributed space….
• Explaining Parallel programming model for Graphs
• Pregel: System that changed graph processing
• Giraph and GraphX
• GraphX Hands on demonstrations:
• Building a Graph: hands on using GraphX
• Building a degree Histogram : hands on using GraphX
• Network Connectedness and Clustering Components : hands on using
• Joining Graph Datasets : hands on using GraphX