CHAT 2016 - China Hotel And Tourism Conference Presentation by Stefan Tweraser. SnapShot Hotel Analytics CEO Stefan Tweraser explains how business leaders can make the most out of the data available to them.
1. How to Make the
Most out of
Enterprise Data?
Dr. Stefan Tweraser
2. The Big Data explosion has fundamentally
changed corporate decision making.
• Ever more data to create information
• Always more stakeholders with access
• Better and real-time situational awareness
• Leaders not able to keep up with these challenges
4. Your company needs to have the
infrastructure to treat data as a utility.
• Utility: useful, profitable, and beneficial
• Infrastructure is more than just computers
• Quality of data drives quality of decision making
• Example: Match-and-merge of customer data
5. Even (or especially) senior leaders need to
learn the fundamentals of analytics.
Recognize
problem or
question
Act
Think
Review
previous
findings
Model solution
and select
variables
Present analysis
and act on
results
Analyze
data
Collect
data
Recognize
problem or
question
8. As strange as it sounds, senior leaders
need to learn how machines learn.
• Machine learning provides computers with the ability
to learn without being explicitly programmed
• Three steps towards “forward looking analytics”
• Understand what happened (descriptive analytics)
• Explore why it happened (diagnostic analytics)
• Predict what is likely to happen next (predictive analytics)
9. A good CEO gets his hands „dirty“ with data,
analytics and the right kind of leadership.
1. Just adding a Chief Data Officer isn’t enough.
2. Analytics is harder than many executives expect:
2 out of 3 analytics initiatives fail .
3. Analytics can spark power shifts in the C-suite -
one of the most difficult challenges for CEOs.
10. Big Data in hospitality industry needs work
but will drive success and guest satisfaction.
Treat your data
as a utility
Keep up with
your quants
Learn how
machines learn
Get your hands
“dirty”
• Too many spreadsheets
• Too many systems
• Not enough benefits
• Too few quants to start with
• Not enough focus
• Not enough knowledge
• Too much manual data
handling
• Not enough focus
• Very rarely top management
focus; little awareness
• Introduce smart BI as first
step, align with SOPs
• (Re)Take analytics 101
• Empower your smartest 10%
• Bring in ‘real” quants
• See what it takes to ‘prohibit’
manual interference w/ data
• Create leadership awareness
• Reward enquiry
• Make it a personal priority
TODAY TO DO
Every day, there are more options to collect, and analyze structured as well as unstructured data to create information.
More stakeholders can instantly access, and share, actionable information.
Real-time situational awareness dramatically increases.
But the managerial and operational ability to act on that data-driven information may not be able to keep up with these challenges.
Utility: the state of being useful, profitable, and beneficial.
Infrastructure: a large store of data accumulated from a wide range of sources within a company and used to guide management decisions.
Quality: data are fit for their intended uses in operations, decision making and planning.
Example: Match-and-merge of customer data: Enterprises have always relied on data to be successful. If the data is so fundamental to the business, it is not surprising that a lot of effort goes into acquiring and handling data and making it available to those who need it. In the process, the data is moved around, manipulated, and consolidated. One of the biggest stumbling blocks in today’s environment, apart from just really bad data, is the matching and merging of disjoint data elements. Being the data acquired through mergers and acquisitions or by consolidating systems, many organizations face the problem of making sense of duplicate (or not) data. Matching and merging data is rarely a purley deterministc effort, more often than not probabilistic matching is required to exploit the statistical probability that a match on particular items is more or less likely to indicate that the records are the same.
What was the source of your data?
How well do the sample data represent the population?
Does your data distribution include outliers? How did they affect the results?
What assumptions are behind your analysis? Might certain conditions render your assumptions and your model invalid?
Why did you decide on that particular analytical approach? What alternatives did you consider?
How likely is it that the independent variables are actually causing the changes in the dependent variable? Might other analyses establish causality more clearly?
Machine learning provides computers with the ability to learn without being explicitly programmed.
Machine-learning software identifies hidden patterns in data and uses those patterns both to group similar data and to make predictions.
Each time new data are added and analyzed, the software gains a clearer view of data patterns and gets closer to making the optimal prediction or reaching a meaningful understanding.
Three steps towards “forward looking analytics”
Understand what happened (descriptive analytics)
Explore why it happened (diagnostic analytics)
Predict what is likely to happen next (predictive analytics)
But don’t act like a machine just because you understand how it learns!