bloom agency at the chief analytics officer forum, europe
TRANSCRIPT
A N A L Y T I C S A S T H E D R I V E R O F V A L U E
P E T E R L A F L I N
P E T E R . L A F L I N @ B L O O M A G E N C Y . C O . U K
@ S Y S T E M S P E T E R
We’re falling into the trap of talking about a number of different
disciplines and roles with the same term, whilst failing to recognise
other things as being data science.
We talk about the same things in different ways, and different things
in the same way.
The importance of language
The Process?
Am I a scientist if I:
• Make random observations about the
weather?
• Collect some weather data?
• Use an IoT device to collect detailed
weather data?
• Show someone a graph of the weather
data I’ve collected?
• Use my data to produce a basic model of
how weather changes near my house?
• Combine my data to provide a robust
abstraction of the laws of “weather”?
• Mathematical Proof
• Mathematical Modeling
• Statistical Interference
• Heuristics
The Scientific Process
With science, we start with a hypothesis and set up experiments to generate
data to test our hypothesis.
With data science, we could take “all of the data” and find reasons to prove or
disprove anything.
W H E R E D O E S D A T A
S C I E N C E F I T ?
If we gave an infinite number of monkeys an infinite number of typewriters,
we’d surely end up with the script to Hamlet.
Emile Borel - 1913
Different Sectors – Different Goals
Public Sector / Government
Transparency, Efficiency, Accountability, Linkage
Value for Citizens from Government
Commerce
Shifting from being service led to being led by the user
Balancing Value for Users & Shareholders
Academia
Looking for global truths, rather than local results
Providing long term foundations to sustain value
Even before academic mathematicians became aware of [the need for data
sciences] the genie was already out of the bottle.
Many companies and institutions simply could not wait for the mathematical
research community to catch up with the applications.
The solutions to disruptive challenges and the novel opportunities
so created were simply too valuable. P. Grindrod – The Mathematical Underpinnings of Analytics 2015
SCIENCE
Data Science process
Facts – which
everyone should agree
on, including the
assumptions on which
they are based.
Conclusions –
which may differ
depending on the
context of the
analysis being
performed.
Strategy – where
the insight is mixed
with domain info to
decide the next
steps.
We need a consistent, robust, formal system that allows us to combine facts
in logically consistent ways across domains, businesses and problem sets.
How do we ensure that facts from different problems, subject to different
assumptions, approximations, models and veracity models, can be combined
effectively in a way which minimises error?
How do we build a data science proof?