RapidMiner Wisdom 2016 - Hortonworks

23
#1 Agile Predictive Analytics Platform for Today’s Modern Analysts Vamsi Chemitiganti General Manager – Financial Services How Predictive Analytics & Big Data are Disrupting Financial Services

Transcript of RapidMiner Wisdom 2016 - Hortonworks

Page 1: RapidMiner Wisdom 2016 - Hortonworks

#1 Agile Predictive Analytics Platform for Today’s Modern Analysts

Vamsi Chemitiganti

General Manager – Financial Services

How Predictive Analytics & Big Data are Disrupting Financial Services

Page 2: RapidMiner Wisdom 2016 - Hortonworks

©2016 RapidMiner, Inc. All rights reserved. - 2 -- 2 -©2016 RapidMiner, Inc. All rights reserved.

State of Global Banking

Page 3: RapidMiner Wisdom 2016 - Hortonworks

©2016 RapidMiner, Inc. All rights reserved. - 3 -

Financial Services and Big Data

Technology vectors

• Cloud computing (OpenStack)

• DevOps and PaaS

• Mobility

• Big Data and analytics

• BPM and microservices

• Software-defined datacenters

Business vectors

• Regulation and risk management

• Compliance and regulation

• Trading systems

• Omni-channel wealth management

• Payments systems

• Bank 3.0

Digital BankBank 3.0s

Focused around business and technology vectors

Page 4: RapidMiner Wisdom 2016 - Hortonworks

©2016 RapidMiner, Inc. All rights reserved. - 4 -- 4 -©2016 RapidMiner, Inc. All rights reserved.

Areas of Impact

Page 5: RapidMiner Wisdom 2016 - Hortonworks

©2016 RapidMiner, Inc. All rights reserved. - 5 -

Key areas within the financial services industry

Page 6: RapidMiner Wisdom 2016 - Hortonworks

©2016 RapidMiner, Inc. All rights reserved. - 6 -

Lifecycle of Big Data adoption

HDP helps FSIs drive efficiency gain..

Page 7: RapidMiner Wisdom 2016 - Hortonworks

©2016 RapidMiner, Inc. All rights reserved. - 7 -

Predictive Analytics on Hadoop

Write data prep and predictive analytics code for Hadoop

It’s complex, requires programming and specialized knowledge of each Hadoop

technology

Push automatically generated computations into Hadoop

It’s code-free, speaks Hadoop for you, and is 10 – 40 x faster to

implement Which would you prefer?

Page 8: RapidMiner Wisdom 2016 - Hortonworks

©2016 RapidMiner, Inc. All rights reserved. - 8 -

Impact of RapidMiner & Data Science

A representative sample only…

Survey of ML algorithms used (stated briefly for confidentiality purposes)

• Classification & Class Probability Estimation

• Regression

• Similarity Matching

• Clustering

• Co-Occurence Grouping

• Profiling

• Link Prediction

• Causal Modeling

• Most use cases typically revolve around a single view of Entity

Page 9: RapidMiner Wisdom 2016 - Hortonworks

©2016 RapidMiner, Inc. All rights reserved. - 9 -- 9 -©2016 RapidMiner, Inc. All rights reserved.

Digital Transformation

Page 10: RapidMiner Wisdom 2016 - Hortonworks

©2016 RapidMiner, Inc. All rights reserved. - 10 -

The digital journey in banking

Page 11: RapidMiner Wisdom 2016 - Hortonworks

©2016 RapidMiner, Inc. All rights reserved. - 11 -- 11 -©2016 RapidMiner, Inc. All rights reserved.

Cyber Security

Page 12: RapidMiner Wisdom 2016 - Hortonworks

©2016 RapidMiner, Inc. All rights reserved. - 12 -

Cyber security

Page 13: RapidMiner Wisdom 2016 - Hortonworks

©2016 RapidMiner, Inc. All rights reserved. - 13 -- 13 -©2016 RapidMiner, Inc. All rights reserved.

Customer Segmentation

Page 14: RapidMiner Wisdom 2016 - Hortonworks

©2016 RapidMiner, Inc. All rights reserved. - 14 -

Customer segmentation process

Page 15: RapidMiner Wisdom 2016 - Hortonworks

©2016 RapidMiner, Inc. All rights reserved. - 15 -- 15 -©2016 RapidMiner, Inc. All rights reserved.

Regulatory Risk Management

Page 16: RapidMiner Wisdom 2016 - Hortonworks

©2016 RapidMiner, Inc. All rights reserved. - 16 -

Proposed SolutionHortonworks Data Platform

LANDING DATA ZONE

L0

STANDARDIZED DATA ZONE

L1

CANONICAL DATA ZONE

L2 RegulatoryReports

Internal Reports

External Reports

Search

REPORTING/ANALYTICS ZONE

L3

Golden Source & Feeds

Master Data

Contrats

Balances

Transaction

Positions

Factors/Scenarios

Market Data

Unstructured Data(hdfs)

Original Data(hdfs)

RAW Datahdfs)

Standardized Data

(Hive/orc)

Materialized View

(Hive/orc)

sqoop/hadoop fs/nfs

Kafka/Storm

Java/Scala

Standardized Data

(Hive/Orc)

Hive/Spark/Scala

Hive/Spark/Scala

MHive/Spark/Scal

Hive/Spark/Scala

Standardized Data

(Hive/Orc)

Hive/Spark/Scala

Hive/Spark/Scala

Hive/Spark/Scala

TBD??

CanonicalPosition

Data(hive/orc)

CanonicalTransaction

Data(Hive/orc)

Hive/Spark

Scala/Python/R etc

Scala/Java

Hive/Spark

ScenariosResults

(Hive/orc)

Data Aggregations

(Hive/orc/Hbase)

Analytics/Reports

(Hive/orc/HBase)

Revision History

(Hive/orc)

Common Repositories/Meta Data Management

Security

Apache Atlas/Falcon/ Custom Solution

Apache Ranger/ Atlas and Custom/Partner Solution

Page 17: RapidMiner Wisdom 2016 - Hortonworks

©2016 RapidMiner, Inc. All rights reserved. - 17 -- 17 -©2016 RapidMiner, Inc. All rights reserved.

AML Compliance

Page 18: RapidMiner Wisdom 2016 - Hortonworks

©2016 RapidMiner, Inc. All rights reserved. - 18 -

Fraud/AML/Compliance Reference architecture

Page 19: RapidMiner Wisdom 2016 - Hortonworks

©2016 RapidMiner, Inc. All rights reserved. - 19 -- 19 -©2016 RapidMiner, Inc. All rights reserved.

Fraud Monitoring &

Detection

Page 20: RapidMiner Wisdom 2016 - Hortonworks

©2016 RapidMiner, Inc. All rights reserved. - 20 -

Fraud DetectionReference architecture

Page 21: RapidMiner Wisdom 2016 - Hortonworks

©2016 RapidMiner, Inc. All rights reserved. - 21 -

Modern Data Architecture with HWX and RM

Page 22: RapidMiner Wisdom 2016 - Hortonworks

©2016 RapidMiner, Inc. All rights reserved. - 22 -

RapidMiner™ Radoop

Big Data Predictive AnalyticsExtends RapidMiner’s visual predictive analytics to Hadoop and Spark

• We speak Hadoop so you don’t have toTranslates predictive analytics into native Hive, MapReduce, Spark, Pig and Mahout – you concentrate on competitive analytics, not Hadoop programming

• COMPLETE insights into your Big DataPushes analytic instructions into Hadoop for computation, so you can analyze the full breadth and variety of your Big Data

– Structured and non-structured

• Not just drag & drop: use your favorite Hadoop scripts, too!Incorporates your favorite SparkR, PySpark, Pig and HiveQLscripts within your predictive analytics workflow

• Safe and sound Integrates with Kerberos authentication, supports data access authorization for Apache Sentry and Apache Ranger – seamless for users, easy admin for IT

Page 23: RapidMiner Wisdom 2016 - Hortonworks

- 24 -CONFIDENTIAL

#1 Agile Predictive Analytics Platform for Today’s Modern Analysts

- 24 -©2016 RapidMiner, Inc. All rights reserved.

Vamsi ChemitigantiGeneral Manager Financial [email protected]