A Statistician's `Big Tent' View on Big Data and Data Science in Health Sciences (Version 11)

A Statistician’s ‘Big Tent’ View on
Big Data and Data Science in
Health Sciences
(Version 11 as of 14.04.2016)
Prof. Dr. Diego Kuonen, CStat PStat CSci
Statoo Consulting
Morgenstrasse 129, 3018 Berne, Switzerland
@DiegoKuonen + kuonen@statoo.com + www.statoo.info
‘Nestl´e Institute of Health Sciences’, Lausanne, Switzerland — April 18, 2016

Abstract
There is no question that big data have hit the business, govern-
ment and scientiﬁc sectors, as well as health sciences. The de-
mand for skills in data science is unprecedented in sectors where
value, competitiveness and eﬃciency are driven by data. How-
ever, there is plenty of misleading hype around the terms ‘big
data’ and ‘data science’. This presentation gives a professional
statistician’s ‘big tent’ view on these terms in health sciences, il-
lustrates the connection between data science and statistics, and
highlights some challenges and opportunities from a statistical
perspective.

‘Statistics has contributed much to data analysis. In
the future it can, and in my view should, contribute
much more. For such contributions to exist, and be
valuable, it is not necessary that they be direct. They
need not provide new techniques, or better tables for
old techniques, in order to influence the practice of
data analysis.’
John W. Tukey, 1962

About myself (about.me/DiegoKuonen)
PhD in Statistics, Swiss Federal Institute of Technology (EPFL), Lausanne, Switzerland.
MSc in Mathematics, EPFL, Lausanne, Switzerland.
• CStat (‘Chartered Statistician’), Royal Statistical Society, United Kingdom.
• PStat (‘Accredited Professional Statistician’), American Statistical Association, United
States of America.
• CSci (‘Chartered Scientist’), Science Council, United Kingdom.
• Elected Member, International Statistical Institute, Netherlands.
• Senior Member, American Society for Quality, United States of America.
• President of the Swiss Statistical Society (2009-2015).
CEO & CAO, Statoo Consulting, Switzerland.
Adjunct Professor of Data Science, Geneva School of Economics and Management
(GSEM), University of Geneva, Switzerland.
Copyright c 2001–2016, Statoo Consulting, Switzerland. All rights reserved.
2

About Statoo Consulting (www.statoo.info)
• Founded Statoo Consulting in 2001.
• Statoo Consulting is a software-vendor independent Swiss consulting ﬁrm spe-
cialised in statistical consulting and training, data analysis, data mining and big
data analytics services.
• Statoo Consulting oﬀers consulting and training in statistical thinking, statistics,
data mining and big data analytics in English, French and German.
Are you drowning in uncertainty and starving for knowledge?
Have you ever been Statooed?
4

5

‘Normality is a myth; there never was, and never will
be, a normal distribution.’
Robert C. Geary, 1947
6

Contents
Contents 7
1. Demystifying the ‘big data’ hype 9
2. Data-driven decision making 22
3. Demystifying the ‘data science’ hype 24
4. What distinguishes data science from statistics? 27
5. Conclusion and opportunities (not only for statisticians!) 32
7

‘Data is arguably the most important natural resource
of this century. ... Big data is big news just about ev-
erywhere you go these days. Here in Texas, everything
is big, so we just call it data.’
Michael Dell, 2014
8

1. Demystifying the ‘big data’ hype
• ‘Big data’ have hit the business, government and scientific sectors.
The term ‘big data’ — coined in 1997 by two researchers at the NASA — has
acquired the trappings of religion.
• But, what exactly are ‘big data’?
The term ‘big data’ applies to an accumulation of data that can not be
processed or handled using traditional data management processes or tools.
Big data are a data management infrastructure which should ensure that the
underlying hardware, software and architecture have the ability to enable ‘learning
from data’, i.e. ‘analytics’.
9

• The following characteristics — ‘the four Vs’ — provide a definition:
– ‘Volume’ : ‘data at rest’, i.e. the amount of data ( ‘data explosion problem’),
with respect to the number of observations ( ‘size’ of the data), but also with
respect to the number of variables ( ‘dimensionality’ of the data);
– ‘Variety’ : ‘data in many forms’, ‘mixed data’ or ‘broad data’, i.e. different
types of data (e.g. structured, semi-structured and unstructured, e.g. log files,
text, web or multimedia data such as images, videos, audio), data sources (e.g.
internal, external, open, public), data resolutions (e.g. measurement scales and
aggregation levels) and data granularities;
– ‘Velocity’ : ‘data in motion’ or ‘fast data’, i.e. the speed by which data are
generated and need to be handled (e.g. streaming data from machines, sensors
and social data);
– ‘Veracity’ : ‘data in doubt’, i.e. the varying levels of noise and processing errors,
including the reliability (‘quality over time’), capability and validity of the data.
10

What do big data look like?
Source: The Association of the British Pharmaceutical Industry (2013). Big Data Road Map
(www.abpi.org.uk/our-work/library/industry/Pages/big-data-road-map.aspx).
11

‘The new medical data ecosystem’
Medicine has entered its data age
( digital revolution).
Medical data are being captured
today from many sources.
Pulling them together and study-
ing what they mean is the next chal-
lenge.
Source: MIT Technology Review (2014). Data-
Driven Health Care (goo.gl/26jQHk).
12

‘As an extension of the digital revolution, the ‘Internet
of Things’ oﬀers particular relevance to health, namely
‘Digital Health’. For example, monitoring people ac-
tively (e.g. via connected wearable tech devices) and
passively (e.g. via stationary sensors) can provide in-
sights into the activity and health of consumers and
patients alike.’
Paul Sonnier, 2016
13

14

• ‘Volume’ is often the least important issue: it is definitely not a requirement to
have a minimum of a petabyte of data, say.
Bigger challenges are ‘variety’ (e.g. combining different omics data generated by
various high-throughput technologies) and ‘velocity’, and most important is ‘veracity’
and the related quality of the data .
Indeed, big data come with the data quality and data governance challenges of
‘small’ data along with new challenges of its own!
• The above definition of big data is vulnerable to the criticism of sceptics that these
four Vs have always been there.
Nevertheless, the definition provides a clear and concise framework to communi-
cate about how to solve different data processing challenges.
But, what is new?
15

‘Scientists have long known that data could create new
knowledge but now the rest of the world, including gov-
ernment and management in particular, has realised
that data can create value.’
Sean Patrick Murphy, 2013
Source: interview with Sean Patrick Murphy, a former senior scientist at Johns Hopkins University
Applied Physics Laboratory, in the Big Data Innovation Magazine, September 2013.
16

‘To get the full business value from big data, compa-
nies need to focus less on the three Vs of big data
(volume, variety, velocity) and more on the four Ms of
big data: ‘Make Me More Money’!’
Bill Schmarzo, March 2, 2015
17

‘Do not focus on the ‘bigness’ of the data, but on the
value creation from the data.’
Stephen Brobst, August 7, 2015
The 5th V of big data: ‘Value’.
18

‘Let us replace the Vs of big data with a single D for
‘Digitalization’.’
Mathias Golombek, October 2, 2015
19

‘It is called big data, but what it really can be is big
success.’
Michael Dell, October 21, 2015
20

‘Data are not taken for museum purposes; they are
taken as a basis for doing something. If nothing is to
be done with the data, then there is no use in collecting
any. The ultimate purpose of taking data is to pro-
vide a basis for action or a recommendation for action.’
W. Edwards Deming, 1942
21

2. Data-driven decision making
• Data-driven decision making refers to the practice of basing decisions on the anal-
ysis of data (i.e. ‘learning from data’), rather than purely on gut feeling and intuition:
22

‘One by one, the various crises which the world faces
become more obvious and the need for hard facts [facts
by analyzing data] on which to take sensible action be-
comes inescapable.’
George E. P. Box, 1976
23

3. Demystifying the ‘data science’ hype
• The demand for ‘data scientists’ — the ‘magicians of the big data era’ — is
unprecedented in sectors where value, competitiveness and eﬃciency are data-driven.
The term ‘data science’ was originally coined in 1998 by a statistician.
Data science — a rebranding of ‘data mining’ — is the non-trivial pro-
cess of identifying valid (that is, the patterns hold in general, i.e. being
valid on new data in the face of uncertainty), novel, potentially useful and
ultimately understandable patterns or structures or models or trends or re-
lationships in data to enable data-driven decision making.
24

• Is data science ‘statistical d´ej`a vu’?
But, what is ‘statistics’?
Statistics is the science of ‘learning from data’ (or of making sense out
of data), and of measuring, controlling and communicating uncertainty.
It is a process that includes everything from planning for the collection of data and
subsequent data management to end-of-the-line activities such as drawing conclusions
of data and presentation of results.
Uncertainty is measured in units of probability, which is the currency of statistics.
Statistics is concerned with the study of data-driven decision making in the face
of uncertainty.
25

‘Statistics has been the most successful information
science. Those who ignore statistics are condemned
to re-invent it.’
Brad Efron, 1997
26

4. What distinguishes data science from statistics?
• Statistics traditionally is concerned with analysing primary (e.g. experimental) data
that have been collected to explain and check the validity of specific existing ideas
(hypotheses).
Primary data analysis or top-down (explanatory and confirmatory) analysis.
‘Idea (hypothesis) evaluation or testing’ .
• Data science (or data mining), on the other hand, typically is concerned with
analysing secondary (e.g. observational or ‘found’) data that have been collected
for other reasons (and not ‘under control’ of the investigator) to create new ideas
(hypotheses).
Secondary data analysis or bottom-up (exploratory and predictive) analysis.
‘Idea (hypothesis) generation’ .
27

• The two approaches of ‘learning from data’ are complementary and should proceed
side by side — in order to enable proper data-driven decision making.
Example. When historical data are available the idea to be generated from a bottom-
up analysis (e.g. using a mixture of so-called ‘ensemble techniques’) could be
‘which are the most important (from a predictive point of view) factors
(among a ‘large’ list of candidate factors) that impact a given process out-
put (or a given indicator)?’.
Mixed with subject-matter knowledge this idea could result in a list of a ‘small’
number of factors (i.e. ‘the critical ones’).
The conﬁrmatory tools of top-down analysis (statistical ‘Design Of Experiments’,
DOE, in most of the cases) could then be used to conﬁrm and evaluate this idea.
By doing this, new data will be collected (about ‘all’ factors) and a bottom-up
analysis could be applied again — letting the data suggest new ideas to test.
28

‘Neither exploratory nor confirmatory is sufficient alone.
To try to replace either by the other is madness. We
need them both.’
John W. Tukey, 1980
29

Data-driven decision making and scientiﬁc investigation (Box, 1976)
Source: Box, G. E. P. (1976). Science and statistics. Journal of the American Statistical Association, 71, 791–799.
30

‘Experiments may be conducted sequentially so that
each set may be designed using the knowledge gained
from the previous sets.’
George E. P. Box and K. B. Wilson, 1951
Scientiﬁc investigation is a sequential learning process!
Statistical methods allow investigators to accumulate knowledge!
31

5. Conclusion and opportunities (not only for statisticians!)
• Decision making that was once based on hunches and intuition should be driven by
data ( data-driven decision making, i.e. muting the HIPPOs).
• Despite an awful lot of marketing hype, big data are here to stay and will over
time, impact every single domain, including health sciences!
The ‘age of big data’ could (and will hopefully) be a new golden era for statistics.
• Statistical principles and rigour are necessary to justify the inferential leap from
data to knowledge.
At the heart of extracting value from big data lies statistics!
32

‘We are in the era of big data, and big data needs
statisticians to make sense of it.’
Eric Schmidt and Jonathan Rosenberg, 2014
Source: Eric Schmidt, Google’s chairman and former CEO, and Jonathan Rosenberg, Google’s
former senior vice president of product, in their 2014 book How Google Works
(New York, NY: Grand Central Publishing — HowGoogleWorks.net).
33

• Lack of expertise in statistics can lead (and has already led) to fundamental errors!
Large amounts of data plus sophisticated analytics do not guarantee success!
Historical results do not guarantee future performance!
• The key elements for a successful (big) data analytics and data science future are
statistical principles and rigour of humans!
• Statistics, (big) data analytics and data science are aids to thinking and not re-
placements for it!
34

• Do not neglect the following four principles that ensure successful outcomes:
– use of sequential approaches to problem solving and improvement, as studies
are rarely completed with a single data set but typically require the sequential
analysis of several data sets over time;
– having a strategy for the project and for the conduct of the analysis of data (
‘strategic thinking’ );
– carefully considering data quality and how data will be analysed ( ‘data pedigree’ );
and
– applying sound subject matter knowledge (‘domain knowledge’), which should
be used to help deﬁne the problem, to assess the data pedigree, to guide analysis
and to interpret the results.
35

‘The data may not contain the answer. The combina-
tion of some data and an aching desire for an answer
does not ensure that a reasonable answer can be ex-
tracted from a given body of data.’
John W. Tukey, 1986
36

• Some challenges from a statistical perspective include
– the ethics of using and linking (big) data, particularly in relation to personal data,
i.e. ethical issues related to privacy ( ‘information rules’ need to be deﬁned),
confidentiality (of shared private information), transparency (e.g. of data uses
and data users) and identity (i.e. data should not compromise identity);
– the provenance of the data, e.g. the quality of the data — including issues like
omissions, data linkage errors, measurement errors, censoring, missing observa-
tions, atypical observations, missing variables ( ‘omitted variable bias’), the
characteristics and heterogeneity of the sample — big data being ‘only’ a sample
(at a particular time) of a population of interest ( ‘sampling/selection bias’,
i.e. is the sample representative to the population it was designed for?);
37

‘To properly analyze data, we must first understand
the process that produced the data. While many
... take the view that data are innocent until proven
guilty, ... it is more prudent to take the opposite ap-
proach, that data are guilty until proven innocent.’
Roger W. Hoerl, Ronald D. Snee and Richard D. De Veaux, 2014
Source: Hoerl, R. W., Snee, R. D. & De Veaux, R. D. (2014). Applying statistical thinking to ‘big data’
problems. Wiley Interdisciplinary Reviews: Computational Statistics, 6, 222–232.
38

– the visualisation of the data, e.g. using and developing effective graphical pro-
cedures;
– spurious (false) associations ( ‘coincidence’ increases, i.e. it becomes more
likely, as sample size increases, and as such ‘there are always patterns’) versus
valid causal relationships ( ‘confirmation bias’);
– the identification of (and the controlling for) confounding factors ( ‘confound-
edness’, i.e. the attribution of the wrong factors to success and the superficial
learning of observational data);
39

– multiple statistical hypothesis testing for explanation (not prediction!) with tens
of thousands or even millions of tests performed simultaneously, e.g. using high-
throughput technology advances, often with complex dependencies between tests
(e.g. spatial or temporal dependence), and the development of further statisti-
cally valid methods to solve large-scale simultaneous hypothesis testing problems;
– the dimensionality of the data ( ‘curse of dimensionality’, i.e. data become
more ‘sparse’ or spread out as the dimensionality increases), and the related usage
and development of statistically valid strategies for dimensionality reduction, e.g.
using ‘embedded’ variable subset selection methods like ‘ensemble techniques’;
40

– the validity of generalisation ( avoid ‘overﬁtting’, i.e. interpreting an ex-
ploratory analysis as predictive);
– the replicability of findings, i.e. that an independent experiment targeting the
same question(s) will produce consistent results, and the reproducibility of find-
ings, i.e. the ability to recompute results given observed data and knowledge of
the data analysis pipeline;
– the nature of uncertainty (both random and systematic);
41

– the balance of humans and computers.
‘There’s no better way to lose money really quickly
than through automated analytics. So I think we have
to be very careful to not totally step out of the picture
and let things go awry.’
Thomas H. Davenport, April 30, 2014
42

‘Business is not chess; smart machines alone can not
win the game for you. The best that they can do for
you is to augment the strengths of your people.’
Thomas H. Davenport, August 12, 2015
43

‘Big data promises to collect large sets of data and
find associations between genes and diseases. There’s
definitely something useful in the data collected, but
the danger is that we have no clue how to interpret
it. Also, you must remember that all statistically sig-
nificant things are not biologically significant. So, it is
definitely not a panacea.’
Walter Gilbert, September 23, 2014
Source: interview with Walter Gilbert, Nobel Laureate 1980 in Chemistry for his contribution to sequence DNA
(theconversation.com/nobel-laureate-big-data-and-full-genome-analysis-not-all-theyre-cracked-up-to-be-31992).
44

‘Most of my life I went to parties and heard a little
groan when people heard what I did. Now they are all
excited to meet me.’
Robert Tibshirani, 2012
Source: interview with Robert Tibshirani, a statistics professor at Stanford University,
in the New York Times, January 26, 2012.
45

‘President’s Invited Speaker’ at the ‘37th Annual Conference of the International
Society for Clinical Biostatistics’ (ISCB 2016) in Birmingham, United Kingdom:
46

Have you been Statooed?
Prof. Dr. Diego Kuonen, CStat PStat CSci
Statoo Consulting
Morgenstrasse 129
3018 Berne
Switzerland
email kuonen@statoo.com
@DiegoKuonen
web www.statoo.info
/Statoo.Consulting

Copyright c 2001–2016 by Statoo Consulting, Switzerland. All rights reserved.
No part of this presentation may be reprinted, reproduced, stored in, or introduced
into a retrieval system or transmitted, in any form or by any means (electronic, me-
chanical, photocopying, recording, scanning or otherwise), without the prior written
permission of Statoo Consulting, Switzerland.
Warranty: none.
Trademarks: Statoo is a registered trademark of Statoo Consulting, Switzerland.
Other product names, company names, marks, logos and symbols referenced herein
may be trademarks or registered trademarks of their respective owners.
Presentation code: ‘Nestle/NIHS.April18.2016’.
Typesetting: LATEX, version 2 . PDF producer: pdfTEX, version 3.141592-1.40.3-2.2 (Web2C 7.5.6).
Compilation date: 14.04.2016.

A Statistician's `Big Tent' View on Big Data and Data Science in Health Sciences (Version 11)

More Related Content

What's hot (20)

Viewers also liked (7)

Similar to A Statistician's `Big Tent' View on Big Data and Data Science in Health Sciences (Version 11) (12)

Recently uploaded (20)

A Statistician's `Big Tent' View on Big Data and Data Science in Health Sciences (Version 11)