agile lab_bigdata_meetup

16
Mattia Bertorello

Upload: paolo-platter

Post on 22-Jan-2018

211 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Agile Lab_BigData_Meetup

Mattia Bertorello

Page 2: Agile Lab_BigData_Meetup

Who is Agile Lab?

GO BIG (data) or GO HOME

Page 3: Agile Lab_BigData_Meetup

Summary

•Why streaming matters

•Why prediction?

•Streaming architecture

•Spark streaming

•Demo time

Page 4: Agile Lab_BigData_Meetup

Why streaming

matters

Ⓒ2015 Agile Lab S.r.l.

Page 5: Agile Lab_BigData_Meetup

Why streaming matters

Data Big Data

Business Reaction

Batch Analysis

Typical BigData Workflow

Page 6: Agile Lab_BigData_Meetup

Why streaming matters

DataReal Time Processing

FASTER REACTIONS MORE PROFITS

Business Reaction

Streaming BigData Workflow

Page 7: Agile Lab_BigData_Meetup

Why streaming matters

• Fleet Management• Insurance• Recommendation• Etc...

Page 8: Agile Lab_BigData_Meetup

Why prediction?

Ⓒ2015 Agile Lab S.r.l.

Page 9: Agile Lab_BigData_Meetup

Why prediction?

• Rule based categorization and clustering is obsolete

• Pattern discovery

• Adaptation to fast changing data

• Smart thinking: no dummies

• Prediction is more valuable

Page 10: Agile Lab_BigData_Meetup

Streaming

architecture

Ⓒ2015 Agile Lab S.r.l.

Page 11: Agile Lab_BigData_Meetup

Streaming architecture

Ingestion Layer

Processing Layer

Serving Layer

Page 12: Agile Lab_BigData_Meetup

Spark Streaming

Page 13: Agile Lab_BigData_Meetup

Spark Notebook

Use Apache Spark straight from the Browser

Page 14: Agile Lab_BigData_Meetup

Demo time...

Ⓒ2015 Agile Lab S.r.l.

Page 15: Agile Lab_BigData_Meetup

Card transaction analysis

PAN CIFRATO | AMOUNT | DESCRIPTION | TIMESTAMP

Classificazione delle transazioni

online/offline

PAN CIFRATO | AMOUNT | DESCRIPTION | TIMESTAMP | ISONLINE

fraud detection algorithm

SQL aggregation

Generazione di allarmi in tempo reale

Page 16: Agile Lab_BigData_Meetup

We are

hiring...

Ⓒ2015 Agile Lab S.r.l.

[email protected]