data
raw number
no context
meaningless
information
context
some meaning
informative
raw numbers
what is there
what is not there
raw facts
descriptive statistics
a table
knowledge
connections and relationships
more meaning
interesting
trends
patterns
no immediate business value
a chart
wisdom
understanding
most meaning
insightful
insights
operational
day-to-day operation
technical expertise
limited accountability
strategic
long term goals
operational understanding [sic!]
high accountability
useful business value
KPI
benchmark
actionable
application
guide action
decission
improvement
advantage
undestanding the whole ⇄ understanding the parts
pre-data
(0) gather requirements
↳ data ↰
(1) gather data
(2) prepare data
↳ ask ↰
(3) explore data
↳ answer ⤴
(4) analyse data
build meaning
information
knowledge
wisdom
↳ visualise
↳ communicate
(5) communicate results
audience
data story
↳ act
decide
improve
have advantage
What is happening?
techniques
explanatory data analysis (EDA)
explore
What is there?
What are the relationships?
What is not there?
descriptive statistics
summarize
frequency
tendency
variability
outliers
visualise
patterns
cluster analysis
cluster
find optimal number of clusters
statistics
domain knowledge
find the distinctive characteristics of a group
groups
relationships
outliers
hypothesis for future analysis
Why is it happening?
causal relationships
root cause
techniques
root cause
drill-down analysis
root cause analysis (RCA)
discovering contributing factors
define event > collect relevant data > determine contributing factors > find root cause > recommend solution
relationships
correlation analysis
regression analysis
hypothesis testing
causalities
danger of wrong conclusions
suggest solutions
What will happen?
possible outcomes
probabilities
trends
techniques
time series forecasting
classification: assign a category to an observation
outcome: category
regression: assign a continuous variable
outcome: value
How shall one act?
possible action
decision making
process optimisation
risk mitigation
techniques
rule based system
reinforcement learning
scenario analysis
simulation analysis
recommendation engine
decision trees
optimisation
Ask a question at the data, answer it with tables and charts. Make your question:
for you
clear
specific
What is the current situation?
What do I need to know?
What do I want to reach?
unambiguous
answerable
know the strategic business goal
company-level
target
purpose
direction
define the business question
qualitative
situation
intuition
experience
team-level
issue
problem
ask the analytical question
quantitative
data
logic
analysis
analytics-team-level
pick the analytical solution viz. tool
categorize analytical type
select technique with:
What is the desired output?
What type of data can be used in what quantity?
What are restrictions on
… time?
… manpower?
… expertise?
… costs?
answer the analytical question with the analytical solution
answer the business question
communicate results
data story telling
business question ⟹ analytical question ⟹ analytical answer ⟹ business answer
bridging the gap between general business strategies and processes to specific data and statistics (NB in both directions!)
business question
stakeholder interview by the analysist
communicate
ask open questions
paraphase
trust
rapport
summarise
focus
control
clarify business terms
business context
What does the business want to know?
Who will use the result?
business goals
How are the results going to be used?
analytical question
"SMART" methodology
specific
focus
narrow scope
root cause
drill down analyis
root cause analysis (RCA)
breakdown business question
refine
defined variables
measurable
defined variables
numerical values concerning business question
data
analysis
statistics
hypothesis
hypothesis testing
actionable
hypothesis testing
supports decision-making
relevant
connected to business problem
address business question
extract key information from business question
time-bound
specific time frame
clarify technical analytical terms
analytical framework
identify data
decide on analytical solution viz. tool
audience1 ⟺ visualisation1 ⟺ answer1 ⟸ question1 ⟸ data ⟹ question2 ⟹ answer2 ⟺ visualisation2 ⟺ audience2
Use tables and charts to tell a story about your data that ends in a conclusion.
Have the user follow you in your story through the charts.
Offer the user the opportunity to come to the same conclusion.
You can answer different questions about the same data with different tables and charts.
You can target different audiences with the same data.
Make your answer:
for the audience
relevant
actionable