Please enable JavaScript.
Coggle requires JavaScript to display documents.
INTRODUCTION AND MOTIVATION, Data granularities, Qualification for use …
INTRODUCTION AND MOTIVATION
Data-Driven
decision making
Base decisions on analytics,
not on feelings and intuition
Enable to turn knowledge
into appropriate decisions
Analysis
Knowledge
Data
Variation
Decisions
Process
Demystifying the
"big data" hype
Big data
Data mgmt IT infrastructure ensure that
hw, sf enable learning from data
Accumulation of data, that could
not be processed using
traditional data mgmt tools
4 V's
Variety
Types
Sources
Resolutions
Velocity
Volume
Observations
Dimensions
Veracity - Quality,
Trust in Data
Noice and errors
Reliability
Quality over time
Capability
Validity
Value -
usefulness of data
Challenges
Think about the problem
Understanf underlying mechanism
driving processes that
generate the data
Provenance
of data (quality)
Duplications
Msrmt errors
Missing observations
Data linkage errors
Omissions
Atypical observations
Missing variables
Sample characteristics
Ethics of
using data
Privacy
Confidentiality
Transparency
Identity
Balance b/w degrading
data VS retaining
sufficient information
Visualisation
Spirous associations VS
Valid causal relationships
Identification of
confounding factors
Replicability of findings
Nature and sources of variation
Balance of humans and computers
Demystifying the
"Internet of things" hype
IoT is about Data
Analytics of things
Analytics on data
generated by IoT devices
It's a KEY: Execute analytics on
the data-gathering devices
Analytics at the edge/
Edge Analytics => Edge Computing
Moving analytics &
"data quality framework"
to the data
Centralised mgmt of
analytics needed
Rule develpment
and maintenance
Common repository for
analytical models
Transparent
central analytics model
Related analytics model
version management
Concerns
Security
Privacy
Analytics governance
Reliability&Scalabilty
of edge devices
Public trust
Key principles for
analytics' success
4 principles
Sequential approach:
seq analysis of several data sets over time
Data Pedigree:
asses before during and after data analysis
Strategic thinking:
Strategy for conducting data analysis
Subject matter knowledge:
knowing the contex, process and problem
to which analytics will be applied
Digital transformation
Focus on transformation
instead of technology
Talent and Culture are key drivers
The costs of
poor data quality
Consequences
of poor data
Missed opportunities
Negative company image &
reputational damage
Loss of customers
Large financial costs
Low customer satisfaction
Poor data quality affect bottom line
Accurate data reduce costs
"Rule of ten":
It costs 10 times to complete
a work when data are flawed
as it does when they are perfect
Data as
strategic asset
Key strategic asset.
Imperative to ensure:
Veracity
Related data quality
Quality is often
compromised by business
Collection
simply collect data no more efficient
sh/b cleansed at the source
to address problems early
Using (big) data = gathering & integrating from diff sources
Top big data tasks:
Data quality governance
Master data mgmt
Data governance
Policies & practices need to put in place
Joint effort b/w IT & business
Framework, defines and help to implement overall mgmt of quality, integrity, usability ... of data use in (digital) ecosystem
Analytics hierarchy
Need for strong data collection
Strong Data pedigree is a key
Base of pyramid
5 essential elements for
succeeding with data
Organizational capability
Technology
Means to monetize
Defense
Quality data
Analytics as a process
of making soup
Phase 2: Data preparation
Phase 3: Analytics model building
Phase 1: Project definition, understand
the needs, priorities, desires, ressources
Phase 4: Analytics model validation
Phase 0: What you want to do?
Phase 5: Analytics model usage
Beginning and end steps
should be well though out
More and more data from
increasing number of sources
Mobile data
Social data
Transactional data
Locational data
Financial data
Family data
Consumption data
Medical data
Data - given
Capta - taken
Amount of data exploding
Analytics - learning from data
Foundation of
digital transformation
Data granularities
Qualification for use
characteristics of BG
Essential
characteristics
of BG