Big Data Analytics for Internet of
Things and Machine Intelligence
Shamshad Ansari
President & CEO
Accure Inc.
Agenda
• Introduction
• Iot & Big Data
• IoT Reference Architecture
• IoT on Momentum
© Copyright Accure Inc. All rights reserved. www.accureanalytics.com
Year 2020
© Copyright Accure Inc. All rights reserved. www.accureanalytics.com
Big Numbers of Small Things
Year 2020
© Copyright Accure Inc. All rights reserved. www.accureanalytics.com
$7.1
Trillion
$1.9
Trillion
50 Billion Connected Devices | Cisco
IoT Economic Vaule Added| Gartner
IoT Solutions Revenue| IDC
IoT
Processor
Cost Less
Than
$1
Gartner
What is IoT?
© Copyright Accure Inc. All rights reserved. www.accureanalytics.com
• Objects connected to
Internet and
communicating with each
other.
• Any object that can be
turned on and off could be
a part of the IoT.
Business Opportunities
© Copyright Accure Inc. All rights reserved. www.accureanalytics.com
Monitor
OptimizeControl
Improved
performance
Reduced cost
New innovative
products
1% savings across
all industries
$150 Billion
Savings
Source: GE
Is IoT New?
© Copyright Accure Inc. All rights reserved. www.accureanalytics.com
We’ve doing this
for many years
Yes
This is really very
facinating and looks
like IoT will change
the world
If IoT is not new, why is this buzz now?
- Widespread cell phone usage
- Reduced price of sensors
- Cheaper Internet connection
- Improvded technology to handle and make use of data
How Big Data Relates to IoT?
How much data?
© Copyright Accure Inc. All rights reserved. www.accureanalytics.com
Gas turbine comressor blade: 500GB/ day | GE
4.4
44
0.132
4.4
2015 2020
Overall
IoT
Zettabyte
An IoT Sensor generates
about 0.5MB per
variable per day
1000 sensors & 1000 variables = 500GB per day
IoT generates big data
IoT and Big Data
© Copyright Accure Inc. All rights reserved. www.accureanalytics.com
Data Ingestion Data Storage
Data
Processing
Data
Reporting
Action
A new mindset and technology is needed to handle IoT data
IoT Challanges
• Scale
– Size and fast growing data
– Large number of users
• Pace
– Fast pace analysis, command and control and cost
– Skill required to do data science and learn new
platform
• Environment
– Complex IoT infrastructure
– Security and privacy
– Lack of standards in data interoperability
© Copyright Accure Inc. All rights reserved. www.accureanalytics.com
IoT Solutions Until Now
• Complex and expensive
• Cost of failing was huge
• Mostly high-volume, homogenous devices,
where software needs are simple
– For simple things like thermostat, refrigerator etc.
© Copyright Accure Inc. All rights reserved. www.accureanalytics.com
Practical Solution Approach
• Think big. Start small
• Start small
– Experiment, learn, and refine
• Think big
– scale
© Copyright Accure Inc. All rights reserved. www.accureanalytics.com
Big Data IoT Architecture
Requirements
• Be able to handle hardware and software
hetrogenity
• Be able to scale big
• Provide low latency
© Copyright Accure Inc. All rights reserved. www.accureanalytics.com
Momentum as an IoT Platform
© Copyright Accure Inc. All rights reserved. www.accureanalytics.com
Momentum provides a configurable platform to manage IoT
• Per device level security
• Massive scalability – millions of concurrent sensors
• Real-time and batch analytics
• Predictive analytics
• Smart alerts
IoT Reference Architecture
© Copyright Accure Inc. All rights reserved. www.accureanalytics.com
Streaming
Gateway
Adapter
Local
Gateway
Scalable
storage
Batch analytics
Realtime
analytics Visualiza
tion
Working Example:
Smart Street Light Analytics
© Copyright Accure Inc. All rights reserved. www.accureanalytics.com
Environment Internal parts Weather
Temperature Temperature Temperature
Pressure Input voltage Pressure
Humidity Output voltage Humidity
Noise Light intensity Sand storm
Rain
Snow
Fog
External DataSensor Data
Momentum in Smart Street Lights
• Data Ingestion
• Predictive Model Building
• Predictive analytics – batch
• Realtime analytics
© Copyright Accure Inc. All rights reserved. www.accureanalytics.com
Momentum Demo – How to create
Streaming Hub
© Copyright Accure Inc. All rights reserved. www.accureanalytics.com
Predictive Modeling
Feature Engineering
• Set A = weather + holiday + weekday + weekend features for the
predicted day
• Set B = number of vehicles that passed each pole in each of the
previous 12 hours
• Set C = number of vehicles that passed in each of the previous 12
days at the same hour
• Set D = number of vehicles that passed in each of the previous 12
weeks at the same hour and the same day
Traing Sets
• Training set 1: feature set A only
• Training set 2: feature sets A+B
• Training set 3: feature sets A+B+C
• Training set 4: feature sets A+B+C+D
© Copyright Accure Inc. All rights reserved. www.accureanalytics.com
Load Data to Momentum
© Copyright Accure Inc. All rights reserved. www.accureanalytics.com
-Specify your source data
- RDBMS
- NoSQL
- Delimited files
- Full or continuos/incremental mode
Pre-processing & Transformation
© Copyright Accure Inc. All rights reserved. www.accureanalytics.com
Provide transformation logic in the form of familiar SQL statement
Predictive Modeling
© Copyright Accure Inc. All rights reserved. www.accureanalytics.com
Predictive Modeling without any programming
Scoring
© Copyright Accure Inc. All rights reserved. www.accureanalytics.com
Model A+B+C is preferred as it has the lowest RMS
Conclusion
• Momentum provide highly scalable IoT
platform
• Big data analytics without any programming
• Reduce implementation time and cost
© Copyright Accure Inc. All rights reserved. www.accureanalytics.com
Next Step
Visit us at www.accureanalytics.com
For an online demo
email us at info@accureanalytics.com
© Copyright Accure Inc. All rights reserved. www.accureanalytics.com

Momentum in Big Data, IoT and Machine Intelligence

  • 1.
    Big Data Analyticsfor Internet of Things and Machine Intelligence Shamshad Ansari President & CEO Accure Inc.
  • 2.
    Agenda • Introduction • Iot& Big Data • IoT Reference Architecture • IoT on Momentum © Copyright Accure Inc. All rights reserved. www.accureanalytics.com
  • 3.
    Year 2020 © CopyrightAccure Inc. All rights reserved. www.accureanalytics.com
  • 4.
    Big Numbers ofSmall Things Year 2020 © Copyright Accure Inc. All rights reserved. www.accureanalytics.com $7.1 Trillion $1.9 Trillion 50 Billion Connected Devices | Cisco IoT Economic Vaule Added| Gartner IoT Solutions Revenue| IDC IoT Processor Cost Less Than $1 Gartner
  • 5.
    What is IoT? ©Copyright Accure Inc. All rights reserved. www.accureanalytics.com • Objects connected to Internet and communicating with each other. • Any object that can be turned on and off could be a part of the IoT.
  • 6.
    Business Opportunities © CopyrightAccure Inc. All rights reserved. www.accureanalytics.com Monitor OptimizeControl Improved performance Reduced cost New innovative products 1% savings across all industries $150 Billion Savings Source: GE
  • 7.
    Is IoT New? ©Copyright Accure Inc. All rights reserved. www.accureanalytics.com We’ve doing this for many years Yes This is really very facinating and looks like IoT will change the world If IoT is not new, why is this buzz now? - Widespread cell phone usage - Reduced price of sensors - Cheaper Internet connection - Improvded technology to handle and make use of data
  • 8.
    How Big DataRelates to IoT? How much data? © Copyright Accure Inc. All rights reserved. www.accureanalytics.com Gas turbine comressor blade: 500GB/ day | GE 4.4 44 0.132 4.4 2015 2020 Overall IoT Zettabyte An IoT Sensor generates about 0.5MB per variable per day 1000 sensors & 1000 variables = 500GB per day IoT generates big data
  • 9.
    IoT and BigData © Copyright Accure Inc. All rights reserved. www.accureanalytics.com Data Ingestion Data Storage Data Processing Data Reporting Action A new mindset and technology is needed to handle IoT data
  • 10.
    IoT Challanges • Scale –Size and fast growing data – Large number of users • Pace – Fast pace analysis, command and control and cost – Skill required to do data science and learn new platform • Environment – Complex IoT infrastructure – Security and privacy – Lack of standards in data interoperability © Copyright Accure Inc. All rights reserved. www.accureanalytics.com
  • 11.
    IoT Solutions UntilNow • Complex and expensive • Cost of failing was huge • Mostly high-volume, homogenous devices, where software needs are simple – For simple things like thermostat, refrigerator etc. © Copyright Accure Inc. All rights reserved. www.accureanalytics.com
  • 12.
    Practical Solution Approach •Think big. Start small • Start small – Experiment, learn, and refine • Think big – scale © Copyright Accure Inc. All rights reserved. www.accureanalytics.com
  • 13.
    Big Data IoTArchitecture Requirements • Be able to handle hardware and software hetrogenity • Be able to scale big • Provide low latency © Copyright Accure Inc. All rights reserved. www.accureanalytics.com
  • 14.
    Momentum as anIoT Platform © Copyright Accure Inc. All rights reserved. www.accureanalytics.com Momentum provides a configurable platform to manage IoT • Per device level security • Massive scalability – millions of concurrent sensors • Real-time and batch analytics • Predictive analytics • Smart alerts
  • 15.
    IoT Reference Architecture ©Copyright Accure Inc. All rights reserved. www.accureanalytics.com Streaming Gateway Adapter Local Gateway Scalable storage Batch analytics Realtime analytics Visualiza tion
  • 16.
    Working Example: Smart StreetLight Analytics © Copyright Accure Inc. All rights reserved. www.accureanalytics.com Environment Internal parts Weather Temperature Temperature Temperature Pressure Input voltage Pressure Humidity Output voltage Humidity Noise Light intensity Sand storm Rain Snow Fog External DataSensor Data
  • 17.
    Momentum in SmartStreet Lights • Data Ingestion • Predictive Model Building • Predictive analytics – batch • Realtime analytics © Copyright Accure Inc. All rights reserved. www.accureanalytics.com
  • 18.
    Momentum Demo –How to create Streaming Hub © Copyright Accure Inc. All rights reserved. www.accureanalytics.com
  • 19.
    Predictive Modeling Feature Engineering •Set A = weather + holiday + weekday + weekend features for the predicted day • Set B = number of vehicles that passed each pole in each of the previous 12 hours • Set C = number of vehicles that passed in each of the previous 12 days at the same hour • Set D = number of vehicles that passed in each of the previous 12 weeks at the same hour and the same day Traing Sets • Training set 1: feature set A only • Training set 2: feature sets A+B • Training set 3: feature sets A+B+C • Training set 4: feature sets A+B+C+D © Copyright Accure Inc. All rights reserved. www.accureanalytics.com
  • 20.
    Load Data toMomentum © Copyright Accure Inc. All rights reserved. www.accureanalytics.com -Specify your source data - RDBMS - NoSQL - Delimited files - Full or continuos/incremental mode
  • 21.
    Pre-processing & Transformation ©Copyright Accure Inc. All rights reserved. www.accureanalytics.com Provide transformation logic in the form of familiar SQL statement
  • 22.
    Predictive Modeling © CopyrightAccure Inc. All rights reserved. www.accureanalytics.com Predictive Modeling without any programming
  • 23.
    Scoring © Copyright AccureInc. All rights reserved. www.accureanalytics.com Model A+B+C is preferred as it has the lowest RMS
  • 24.
    Conclusion • Momentum providehighly scalable IoT platform • Big data analytics without any programming • Reduce implementation time and cost © Copyright Accure Inc. All rights reserved. www.accureanalytics.com
  • 25.
    Next Step Visit usat www.accureanalytics.com For an online demo email us at info@accureanalytics.com © Copyright Accure Inc. All rights reserved. www.accureanalytics.com