Data Science as a Service
Dr. Pouria Amirian (Pouria.Amirian@ndm.ox.ac.uk)
Big Data Project Coordinator, The Global Health Network, University of Oxford
Intersection Of Cloud Computing And Data Science
outline
 Data Science
 What data science is
 Steps in a Data Science project
 Experiments
 Using AzureML
 Big Data issues
 In a data science project
 Methods in analysis
2
What is Data Science?
 Practice of obtaining useful insights from data
 3 Vs of Big Data:
 Volume
 Variety
 Velocity
 + other Vs
 It applies to large volume data (volume)
 It applies to semi-structured and unstructured data (variety)
 It sometimes applies to real-time or fast changing data
(velocity)
 It applies to small and traditional static data
3
Data Science as a team sport
4
Math
Statistical Learning
Linguistics
Machine Learning
Signal Processing
Programming
Storage/Data StructureOperations Research
Distributed and High
Performance Computing
Data Science from analytics point of view
 Analytics Spectrum:
5
Descriptive
Diagnostic
Predictive
Prescriptive
What happened?
Why did it happen?
What will happen?
What should I do?
Data Science Vs. Business Intelligence
 Analytics Spectrum:
6
What happened?
Why did it happen?
What will happen?
What should I do?
Traditional BI
Descriptive
Diagnostic
Predictive
Prescriptive
Why is it so popular? why it matters?
 A) More Available and Usable Data
 McKensey: Organizations that use data science to make
decisions are more productive and deliver higher ROI
 Gartner: Organizations that invest in modern data infrastructure
will outperform their peers by up to 20%
7
Why is it so popular? why it matters?
 B) Increased Awareness of Machine Learning Techniques
 A subset of machine learning algorithms are now more widely
understood since they have been tried and tested by early
adopters such as Netflix and Amazon (Recommendation engines).
 while many people may not know details of the algorithms used,
they now increasingly understand their research/business value.
8
Why is it so popular? why it matters?
 C) More Accuracte Analysis
 The large volumes of data being collected also enables you to
build more accurate predictive models.
 The larger sample size, the smaller the margin of error. This in turn
increases the accuracy of predictions from your model.
9
Why is it so popular? why it matters?
 D) Faster and Cheaper Computation
 Today, a smartphone’s processor is up to five times more
powerful than that of a desktop computer 20 years ago.
 Price of computation is decreased
 Capacity of computation is increased
 dramatic gains in technology, productivity, innovations etc.
10
The Data Science Workflow
Problem Definition
Data Collection and
Preparation
Model
Development
Model
Deployment
Performance
Improvement
11
Critical
Very Important
Time Consuming
Fun :D
Iterative
Cumbersome :(
Critical
The Data Science Workflow
Problem Definition
Data Collection and
Preparation
Model
Development
Model
Deployment
Performance
Improvement
12
• Domain Knowledge
• Separation of Concerns
• Prioritize each problem
• Selection or right data
• Data Transformation
• Missing Values
• Exploratory analysis
• Right algorithm
• Test accuracy
• Test other algorithms
• Validate
• Turning data scientist model
to developer code
(R to C#)
• Monitor the performance of
deployed model
• Re-Training model
• Re-Deploying model
• Re-monitoring
The Data Science Workflow
Big Data
Issues (I)
Problem Definition
Data Collection and
Preparation
Model
Development
Model
Deployment
Performance
Improvement
13
Solutions to overcome the big data issues
14
 1- Use advanced research computing
(http://www.arc.ox.ac.uk/)
Solutions to overcome the big data issues
15
 2-Create and use a Hadoop Cluster
 Open source (Apache)
 It is based on two components
HDFS
MapReduce
MapReduce
16
HortonWorks
17
Cloudera
18
MapR
19
open source won't prevent vendor
lock-in!!!
20
Third Solution
 Microsoft’s
Cloud Computing
21
AzureML (Azure Machine Learning)
 Azure ML provides an easy-to-use and powerful set of cloud-
based data transformation and machine learning
tools.
 AzureML Studio (or Studio for short)
 It has many modules for data transformation, analysis,
visualization,…
 It supports R and Python
 It is under heavy development
 www.studio.azureml.net
22
AzureML Workflow
23
Data Input
Data Transformation (Project)
Split Data(training and test)Learning Algorithm
Train the Learning Algorithm
Validate the Algorithm(Score)
Evaluate Model Performance
First Experiment:
Predicting Price of Car
AutomobileFullModuleModel02-03-2015
24
Second Experiment: Using R in ML Studio
AutomobileRTransformation02-03-2015
25
Third experiment: comparing two models
AutomobileFullModuleTwoModel02-03-2015
26
Fourth experiment: Creating Web service
 Very easy just some clicks!!!!
 Make: bmw
 Engine-size: 164
 Horse-power: 121
 highway—mpg: 25
 Its actual price is 24,565
27
Tips
 Data input can come from a variety of data interfaces,
including HTTP connections (any filesharing service like
dropbox, googleDrive, oneDrive), SQLAzure, and Hive Query.
 You can use functionality in all supported R modules (410)
 You can write your utility functions and upload it as another
module
 It is under heavy development
 Two weeks ago the process for web service publication changed
 Two months ago there was no support for Python
 Two months ago around 400 R packages were supported
 …
28
Big Data Issues (II)
 High dimensional data or wide data
 Using various methods needs knowledge of those methods
 Traditional methods are not efficient enough (unstable)
 Least Squares for example
29
Advantages of AzureML
 Solutions can be quickly deployed as web services.
 Models run in a highly scalable cloud environment.
 using the R and Python language for solution-specific
functionality.
 It creates minimum code for consuming the web service
in R and Python (and C#)
 It can be run from anywhere
30
“
”
Big Data is not about Data.
The value in big data is in
Analytics.
GARY KING
Thanks for your attention
Time for Q/A

Data Science as a Service: Intersection of Cloud Computing and Data Science

  • 1.
    Data Science asa Service Dr. Pouria Amirian (Pouria.Amirian@ndm.ox.ac.uk) Big Data Project Coordinator, The Global Health Network, University of Oxford Intersection Of Cloud Computing And Data Science
  • 2.
    outline  Data Science What data science is  Steps in a Data Science project  Experiments  Using AzureML  Big Data issues  In a data science project  Methods in analysis 2
  • 3.
    What is DataScience?  Practice of obtaining useful insights from data  3 Vs of Big Data:  Volume  Variety  Velocity  + other Vs  It applies to large volume data (volume)  It applies to semi-structured and unstructured data (variety)  It sometimes applies to real-time or fast changing data (velocity)  It applies to small and traditional static data 3
  • 4.
    Data Science asa team sport 4 Math Statistical Learning Linguistics Machine Learning Signal Processing Programming Storage/Data StructureOperations Research Distributed and High Performance Computing
  • 5.
    Data Science fromanalytics point of view  Analytics Spectrum: 5 Descriptive Diagnostic Predictive Prescriptive What happened? Why did it happen? What will happen? What should I do?
  • 6.
    Data Science Vs.Business Intelligence  Analytics Spectrum: 6 What happened? Why did it happen? What will happen? What should I do? Traditional BI Descriptive Diagnostic Predictive Prescriptive
  • 7.
    Why is itso popular? why it matters?  A) More Available and Usable Data  McKensey: Organizations that use data science to make decisions are more productive and deliver higher ROI  Gartner: Organizations that invest in modern data infrastructure will outperform their peers by up to 20% 7
  • 8.
    Why is itso popular? why it matters?  B) Increased Awareness of Machine Learning Techniques  A subset of machine learning algorithms are now more widely understood since they have been tried and tested by early adopters such as Netflix and Amazon (Recommendation engines).  while many people may not know details of the algorithms used, they now increasingly understand their research/business value. 8
  • 9.
    Why is itso popular? why it matters?  C) More Accuracte Analysis  The large volumes of data being collected also enables you to build more accurate predictive models.  The larger sample size, the smaller the margin of error. This in turn increases the accuracy of predictions from your model. 9
  • 10.
    Why is itso popular? why it matters?  D) Faster and Cheaper Computation  Today, a smartphone’s processor is up to five times more powerful than that of a desktop computer 20 years ago.  Price of computation is decreased  Capacity of computation is increased  dramatic gains in technology, productivity, innovations etc. 10
  • 11.
    The Data ScienceWorkflow Problem Definition Data Collection and Preparation Model Development Model Deployment Performance Improvement 11 Critical Very Important Time Consuming Fun :D Iterative Cumbersome :( Critical
  • 12.
    The Data ScienceWorkflow Problem Definition Data Collection and Preparation Model Development Model Deployment Performance Improvement 12 • Domain Knowledge • Separation of Concerns • Prioritize each problem • Selection or right data • Data Transformation • Missing Values • Exploratory analysis • Right algorithm • Test accuracy • Test other algorithms • Validate • Turning data scientist model to developer code (R to C#) • Monitor the performance of deployed model • Re-Training model • Re-Deploying model • Re-monitoring
  • 13.
    The Data ScienceWorkflow Big Data Issues (I) Problem Definition Data Collection and Preparation Model Development Model Deployment Performance Improvement 13
  • 14.
    Solutions to overcomethe big data issues 14  1- Use advanced research computing (http://www.arc.ox.ac.uk/)
  • 15.
    Solutions to overcomethe big data issues 15  2-Create and use a Hadoop Cluster  Open source (Apache)  It is based on two components HDFS MapReduce
  • 16.
  • 17.
  • 18.
  • 19.
  • 20.
    open source won'tprevent vendor lock-in!!! 20
  • 21.
  • 22.
    AzureML (Azure MachineLearning)  Azure ML provides an easy-to-use and powerful set of cloud- based data transformation and machine learning tools.  AzureML Studio (or Studio for short)  It has many modules for data transformation, analysis, visualization,…  It supports R and Python  It is under heavy development  www.studio.azureml.net 22
  • 23.
    AzureML Workflow 23 Data Input DataTransformation (Project) Split Data(training and test)Learning Algorithm Train the Learning Algorithm Validate the Algorithm(Score) Evaluate Model Performance
  • 24.
    First Experiment: Predicting Priceof Car AutomobileFullModuleModel02-03-2015 24
  • 25.
    Second Experiment: UsingR in ML Studio AutomobileRTransformation02-03-2015 25
  • 26.
    Third experiment: comparingtwo models AutomobileFullModuleTwoModel02-03-2015 26
  • 27.
    Fourth experiment: CreatingWeb service  Very easy just some clicks!!!!  Make: bmw  Engine-size: 164  Horse-power: 121  highway—mpg: 25  Its actual price is 24,565 27
  • 28.
    Tips  Data inputcan come from a variety of data interfaces, including HTTP connections (any filesharing service like dropbox, googleDrive, oneDrive), SQLAzure, and Hive Query.  You can use functionality in all supported R modules (410)  You can write your utility functions and upload it as another module  It is under heavy development  Two weeks ago the process for web service publication changed  Two months ago there was no support for Python  Two months ago around 400 R packages were supported  … 28
  • 29.
    Big Data Issues(II)  High dimensional data or wide data  Using various methods needs knowledge of those methods  Traditional methods are not efficient enough (unstable)  Least Squares for example 29
  • 30.
    Advantages of AzureML Solutions can be quickly deployed as web services.  Models run in a highly scalable cloud environment.  using the R and Python language for solution-specific functionality.  It creates minimum code for consuming the web service in R and Python (and C#)  It can be run from anywhere 30
  • 31.
    “ ” Big Data isnot about Data. The value in big data is in Analytics. GARY KING Thanks for your attention Time for Q/A