Transcript AGU2014-ED31E-3455_Fox - Tetherless World Constellation

AGUFM14 – ED31E-3455 (MS Hall A-C)
The Landscape – Data Ecosystem and What Makes Up a Data Scientist?
Producers
Consumers
Overused Venn
diagram of the
intersection of
skills needed for
Data Science
(Drew Conway)
Experience
The Anatomy and Physiology of Data Science
Data
Peter Fox1 ([email protected])
http://tw.rpi.edu/web/Courses
Creation
Gathering
(1Rensselaer Polytechnic Institute 110 8th St., Troy, NY, 12180 United
States – see Acknowledgements)
“Data” Science
Lt. Cmdr Data, Star Trek TNG
Data science is advancing the
inductive conduct of science and
is driven by the greater volumes,
complexity and heterogeneity of
data being made available over
the Internet.
Data science combines aspects of
data management, library science,
computer science, and physical
science using supporting
cyberinfrastructure and
information technology.
It is changing the way all of these
disciplines do both their
individual and collaborative
work.
Key methodologies in application
areas based on real research
experience are taught to build a
skill-set.
BigData Science (Data Analytics)
Data and Information analytics extends analysis (descriptive and predictive
models to obtain knowledge from data) by using insight from analyses to
recommend action or to guide and communicate decision-making. Thus,
analytics is not so much concerned with individual analyses or analysis steps,
but with an entire methodology. The world at-large is confronted with
increasingly larger and complex sets of structured/unstructured information;
from sensors, instruments, and generated by computer simulations; data is
"hidden" in websites, application servers, social networks and on mobile
devices. In commerce and industry, analytics-driven enterprises are becoming
mainstream. Yet, there is a shortfall in the key education skills needed to meet
the growing needs. Traditional enterprises are moving toward analytics-driven
approaches for core business functions. In the government and corporations,
cybersecurity problems are prevalent.
Key topics include:
Lt. Cmdr Data and Friends
advanced statistical
computing theory,
multivariate analysis,
and application of
computer science
courses such as data
mining and machine
learning and change
detection by
uncovering
unexpected patterns
in data.
Presentation
Organization
The DataInformationKnowledge
Ecosystem
(Fox; derived)
Knowledge
Integration
Conversation
Context
MOTIVATION
 Whether the science (especially geosciences) community at-large likes it or not,
the co-opting of the term Data Science by the private sector has led to increased
hype over data science as a career and as a means to solve challenging data
problems, and lack of educational innovation in curricula for data science.
 If the full benefits of a new generation of statistical and analytical software tools
that operate on high-performance computational infrastructure are to be attained,
adequate attention to the 'science of data science' is needed. In this contribution,
we present a science view of data science both from an education and research
perspective.
 We introduce a research agenda that explores the key challenges that must be met
to meet the needs of research driven by large-scale data analytics.
 We focus on three, as-yet untapped, data science topics:
 understanding scale in systems,
 sparse systems, and
 abductive reasoning.
 We conclude with a specific call to action to make progress on the
aforementioned topics.
Information
1
 Data Science primarily advances the inductive conduct of
science but to understand scale in systems, accommodate
sparse systems, and provide for abductive reasoning, data
scientists must progress to data analyticists.
Anatomy & Physiology
 Anatomy (as an individual)
 Data Life Cycle – Acquisition,
Curation and Preservation
 Data Management and Products
 Forms of Analysis, Errors and
Uncertainty
 Technical tools and standards
 Physiology (in a group)
 Definition of Science Hypotheses,
Guiding Questions
Anatomy study of the structure and
relationship between body parts
Physiology is the study of the
function of body parts and the body
as a whole.
Learning Outcomes
Through class lectures, practical sessions, written and oral
presentation assignments and projects, students should:
 Develop and demonstrate skill in Data Collection and Data
Management
 Demonstrate proficiency in Data/ Information Product
Generation
 Demonstrate science-driven Analysis and Presentation of
Integrated Datasets from the Web
 Demonstrate the development and application of Data Models
 Convey knowledge of and apply Data and Metadata Standards
and explaining Provenance
 Finding and Integrating Datasets
 Apply Data Life-Cycle principles, construct Data Workflows
 Presenting Analyses and Viz.
 Develop and demonstrate skill in Data Tool Use and
 Presenting Conclusions
Anatomy & Physiology
 Anatomy (individual)
 Intermediate Skill in parametric
and non-parametric statistics
 Application of a broad spectrum
of Data Mining and Machine
Learning Algorithms
 Ability to cross-validate and
optimize models
 Application to specific datasets
 Physiology (term project)
 Definition of Science Hypotheses,
with Prediction/ Prescription Goal
 Cleaning and Preparing Datasets
 Validating and Verifying Models
 Presenting Ideas and Results
Call To Action
 Data Science across the curriculum
 Same as “Calculus”
 And … Intro to Statistics
 Data Management is Second Nature
 Like operating an instrument
 Openness/ sharing is the natural
state
 As-a-whole, the Data Scientist works
collaboratively and is recognized and
rewarded by peers and organizations
Evaluation
Learning Outcomes
 To demonstrate knowledge of relevant analytic methods,
and to recognize and apply quantitative algorithms,
techniques and interpret results
 To demonstrate strategic thinking skills, combined with a
solid technical foundation in data and model-driven
decision-making.
 To develop ability to apply critical and analytical methods to
formulate and solve science, engineering, medical, and
business problems
 Examine real-world examples to place data-mining
techniques in context, develop data-analytic thinking, to
illustrate that their application is art and science.
 Must effectively communicate analytic findings to nonspecialists.
 Must develop and demonstrate a working knowledge of
decision making under uncertainty, be able to build
optimization models that incorporate random parameters:
static stochastic optimization, two-stage optimization with
recourse, chance-constrained optimization, and sequential
decision making.
Call To Action
 Institutions to provide reliable, highfunctionality data infrastructures that
facilitate analytics
 Provision of intermediate to
advanced Statistics to
undergraduates and early graduate
students
 Well-curted datasets are made
widely available along with
developed models and validation
statistics
 All results are under continuous
scrutiny, are traceable and verifiable
Acknowledgments:
Glossary:
Sponsors:
TWC eScience Group
W3C Provenance Working Group
RPI – Rensselaer Polytechnic Institute
TWC – Tetherless World Constellation at Rensselaer Polytechnic Institute
Rensselaer Polytechnic Institute
Tetherless World Constellation