WEF_Data_Operationalization_Principles_Design_2020.pdf

(875 KB) Pobierz
10 Principles of Mobility
Data Operationalization
E N G AG E M E N T M AT E R I A L
JUNE 2020
Cover:
Markus Spiske
Contents
3
4
4
5
6
7
8
8
9
11
12
13
13
1
Before you even get started
Principle 1
Question your assumption
Principle 2
Take inventory, and take it often
Principle 3
There is no need to reinvent the wheel
Principle 4
Data likes (selective) company
2
While you are at it
Principle 5
Data is nothing but insights are everything
Principle 6
Less is most definitely not less
Principle 7
Ambition has many traps
3
Justifying the whole exercise
Principle 8
Pilots are your best friends and your worst enemies
Principle 9
Don’t forget your checks and balances
Principle 10
Big questions are only as good as small realities
Author
Mouchka Heller,
Autonomous and Automotive Mobility,
World Economic Forum
© 2020 World Economic Forum. All rights
reserved. No part of this publication may
be reproduced or transmitted in any form
or by any means, including photocopying
and recording, or by any information
storage and retrieval system.
10 Principles of Mobility Data Operationalization
2
1
Before you even
get started
10 Principles of Mobility Data Operationalization
3
Principle 1
Question your assumptions
Your data is just as subjective as your are
Data is not a magic blanket of objectivity. It
carries with it your own biases, prejudices and
assumptions. For example, most data collection
processes for safety in public transit systems track
formal complaints to the police for crimes such as
rape and assault. Such processes assume that
1) all victims would formally file a complaint; 2)
that all victims would know how to navigate the
process for filing a complaint; and 3) those that
are not victims of crime feel safe on public transit.
If a decision-maker looks at the data they have
collected with the assumption that it is inherently
objective and presents a complete picture of safety
in a given public transit system, that decision-
maker is not getting a holistic picture of safety
nor will they be able to present a holistic picture
of safety to the general public. Many crimes go
unreported, it’s a painstaking process to report a
crime, and even when you do report one, there’s
no resolution. There are other examples of crimes
that are difficult to include. For example, a bus
station defaced with racial slurs may be empty at
night but since there are no formal complaints filed,
public officials may deem that station safe. Such
blind spots can make insights drawn from datasets
as valueless as the subjective assessments that
were meant to be tested in the first place.
Hire people who are different from you
Few things in this world are as human as biases
and it is humans who are building datasets. Within
a team meant to collect and process data to
generate insights for decision-making, make sure
to hire or include colleagues of socioeconomic
diversity who can ideally challenge each other’s
prejudices. Data points are only the reflection of
on-the-ground realities. Having as many pairs of
eyes to test that reflection helps to ensure higher
data quality.
Get ready to be wrong
If you are lucky and do this well, you will generate
insights that will challenge your world view,
irrelevant of how small or large your project
may be. In a project supporting cross-border
commuters, a team recently assumed that nurses
could be a great focus group because of their
unconventional hours and relative inability to work
remotely to ease their weekly pace. However, in
qualitative focus groups where nurses were allowed
to openly share all their concerns, the aggregated
feedback showed that nurses were reluctant to be
included in a shared commuting solution because
of a preference to keep their overtime hours
informal. The entire framework of the operation
had to be rethought because of this single point
of data, which is the whole reason to engage in a
holistic data-driven analysis in the first place.
Principle 2
Take inventory, and take it often
You don’t know what data you have until you look
One assumption you have likely already made is
how much data you do and do not own. Having
a master dataset does not mean you own all
the relevant sub-datasets or that you managed
subscriptions well enough to be sure of its quality.
You also need to pay attention to duplication and
even contradiction. You don’t necessarily know
how much data your colleagues have collected,
how their collection processes have differed, what
can be used in combination and what should be
rejected out of hand.
10 Principles of Mobility Data Operationalization
4
The task of inventory never ends
Just because you have started your process of
looking at what data you own and do not own
does not mean your colleagues have stopped
collecting and erasing data. In some jurisdiction,
privacy laws mean that organizations even have
set up automatic processes to delete data points
after a certain period of inactivity or on request.
Doing inventory manually works, of course, but
best practice would be to set up a framework of
data integration that is updated continuously and
automatically so you know what you have and not
what you had.
Inventory is your most important data
The task of integrating disparate datasets is not
trivial, cheap or short. It infers the need for a
standardization process that can apply across
all datasets, sometimes complex data-sharing
agreements, even for internal projects, and an
interface capable of not only containing but also
visualizing and ideally synthesizing the data. For
projects that necessitate the involvement of non-
technical staff, this integrated tool also needs to
have user-friendly, front-end capabilities, as well
as institutionalized processes for updates and
corrections.
Principle 3
There is no need to reinvent the wheel
Data-exchange frameworks and standards already exist
Numerous cities and organizations around the
world have already tackled the question of data-
exchange frameworks, setting up global standards
that are rapidly gaining in adoption and popularity.
With dedicated full-time staff and enormous
resources, these organizations still recognize the
extraordinary complexity of designing, testing and
catalysing the adoption of common data-exchange
rules. Public-private data-sharing agreements alone
can require full-time work from entire teams for
large amounts of time and can end up looking like
snowflakes. Cities have adopted largely different
stances on data sharing, so the same private
institution also might have different agreements
in place around the world and have to manage a
separate process for the transfer of data across
these regions.
Before diving into your data process and analysis,
leverage the knowledge that already exists and rely
on data-exchange frameworks and standardization
processes that have already been tested and
approved in your particular region. A good example
of a non-profit leader in this field is the
Open
Mobility Foundation.
Data privacy regulations and recommendations already exist
Similarly, do not engage in a guessing game when
it comes to data-privacy settings for your product,
services or project. Privacy is a culturally sensitive
concept that might be completely different from
what you would expect from culture to culture and
region to region. Rather than go with what you think
is right, establishing focus groups to ask others
what they’d prefer, or try and redefine a complex
balance, identify the strictest standard in the regions
in which you operate and set it as your baseline.
It will help with data transfers, adequacy status,
public-private partnerships and public engagement.
Chances are you’re not the first to try this
Of course, your idea is unique. Only it isn’t. There
is a high probability that someone, somewhere,
has tried something similar. If that entity partnered
with a public or civic organization, there is probably
a wealth of information available regarding best
practices and lessons learned. Take time to
research similar initiatives to yours and engage
project managers. No one is invulnerable to a
mistake, misconception, or lapse in judgement.
When dealing with an area as sensitive and
powerful as data, you don’t want to repeat the
mistakes of others.
10 Principles of Mobility Data Operationalization
5

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika:

Zgłoś jeśli naruszono regulamin