the future is an event max

18
The future is an event ADDRESSING MODERN DATA CHALLENGES MAX COTTICA

Upload: lauren-campbell-assoc-cipd

Post on 22-Jan-2018

17 views

Category:

Technology


1 download

TRANSCRIPT

Page 1: The future is an event max

The future is an eventADDRESSING MODERN DATA CHALLENGES

MAX COTTICA

Page 2: The future is an event max

Computer Shop Clerk

(5 years)

IT Development Manager

(10 years)

SQL DBA/Developer

(5 years)

Data Warehouse Developer/Junior Manager

(6 years)

Data Warehouse Manager

(5 years)

Global Data Integration Senior Manager

(4 years)

Head of Data Science

and Solutions

(1 year)

About me

1980s 1990s 2000s 2010s 2017

Page 3: The future is an event max

1980

‘80s

‘90s

‘00s

‘10s“My goal is to use predictive analytics in conjunction with scenario based algorithms to produce prescriptive analytics and actionable events.”

“My goal was to use aggregated reports to produce KPI reports andto use predictive models to estimate future growth.”

“My goal was to aggregate and integrate data and offer analytics

based on different segmentations like time or geography .”

“My goal was to produce ad-hoc reports for a large variety ofdepartments using a centralized media for distribution.”

“My goal was to produce reports.”

TIER 1

TIER 0

TIER 2

TIER 3

Descriptive Analytics

Diagnostic Analytics

Predictive Analytics

Prescriptive Analytics

What happened

Roll-ups and how, when and where

Identify problems and fire alerts

Why and forecasting

What will happen if…Next best action

Max vs. Analytics in the years:

Page 4: The future is an event max

Modern data

Page 5: The future is an event max

Getting the flow right – 1997 to 2017

Capture Store Transform Publish

Modern Data Does not like

batch

Batch it’s all it knows

Page 6: The future is an event max

Getting the flow right - Today

Page 7: The future is an event max

Maybe not a lake but just a river Real customer value is in speed streams

Decisions in real time

Aligning to IoT speed

Challenges and trends:

SVOT vs MVOT

Switching from Lambda to Kappa architecture

Kafka for the win

Speed modelling

OTS model for analytics

Replay and Search Analytics

Real time application

Page 8: The future is an event max

SVOT vs MVOT

Truth

Finance

HR

IT

Sales

IT

Finance

Sales

HR

Truth Truth

Truth Truth

Truth

Page 9: The future is an event max

The future is an event - Lambda

Page 10: The future is an event max

Kafka for the winApache Kafka is an open-source stream processingplatform developed by the Apache Software Foundationwritten in Scala and Java. Originally developed by LinkedIn as a messaging queue system.

Page 11: The future is an event max

The future is an event - Kappa

Page 12: The future is an event max

Speed modelling

Page 13: The future is an event max

Traditional dimensional model

Surrogate Key DoorID DoorLocation DoorStatus Timestamp IsLatestRecord DWHField1 DWHFiled2 DWHField3

1 1 North Closed 12:00:01 N 22/03/1965 123456 userA

2 2 South Open 12:00:02 N 22/03/1965 123456 userA

3 1 North Open 12:00:03 N 22/03/1965 123456 userA

4 1 North Closed 12:00:04 N 22/03/1965 123456 userA

5 2 South Closed 12:00:05 N 22/03/1965 123456 userA

6 2 South Open 12:00:06 N 22/03/1965 123456 userA

7 2 South Closed 12:00:07 N 22/03/1965 123456 userA

8 1 North Open 12:00:08 N 22/03/1965 123456 userA

9 1 North Closed 12:00:09 Y 22/03/1965 123456 userA

10 2 South Open 12:00:11 Y 22/03/1965 123456 userA

Data streams

Page 14: The future is an event max

Micro-dimensions

Data streams

Surrogate Key DoorID DoorLocation DWHField1 DWHFiled2 DWHField3

1 1 North 22/03/1965 123456 userA

2 2 South 22/03/1965 123456 userA

3 1 North 22/03/1965 123456 userA

4 1 North 22/03/1965 123456 userA

5 2 South 22/03/1965 123456 userA

6 2 South 22/03/1965 123456 userA

7 2 South 22/03/1965 123456 userA

8 1 North 22/03/1965 123456 userA

9 1 North 22/03/1965 123456 userA

10 2 South 22/03/1965 123456 userA

DoorID DoorStatus Timestamp

1 Closed 12:00:01

1 Open 12:00:03

1 Closed 12:00:04

1 Open 12:00:08

1 Closed 12:00:09

DoorID DoorStatus Timestamp

2 Open 12:00:02

2 Closed 12:00:05

2 Open 12:00:06

2 Closed 12:00:07

2 Open 12:00:11

Entities or partitions

Page 15: The future is an event max

OTS model for green field analytics

Enterprise

Business driven

Design

12 to 24 months

E2E solution

Small scale

Architecture

Subject driven

6 to 12 months

Qlik / Tableau

Exploration

Discovery

Insight

3 to 6 months

SQL, R, Python

Operational

Tactical

Strategic

CLOUD

Page 16: The future is an event max

Replay, Untapped and Search Analytics Capability to replay past transactions in a segmentations and predictive context

Fine tune algorithms for prescriptive analytics

Anomaly detection techniques to align past to present

Video, Audio and other media

We now have the technology to leverage this data

Needs specific use cases

Using NLP for Google like searches

Results are of various nature from a list to actual insights to graphs

Elastic/Solr will play a big role

Page 17: The future is an event max

CV

OnlineProfiles

Cover Letter

ML

NLP

DL

Job Specs

AIMatches

WEB UI

Email

BOT

Mobile

EmployersEmployees

MicroServices

APIs

BUILDING A REAL TIME SOLUTION (TO DISRUPT THE RECRUITMENT INDUSTRY)

Page 18: The future is an event max

[email protected]

“Oh yes, my company is doing Agile…”

“Does your CEO have a backlog ?”

“Of course not !”

“I have got news for you, you are not doing Agile…”