The document discusses using IBM Watson Analytics to predict the results of the 2016 March Madness college basketball tournament. It introduces Watson Analytics and describes how to prepare basketball tournament data by combining 15 seasons into a single data set. Predictive analysis is conducted on the data to identify the top predictors of teams that reach the Final Four and championship game. A comparison is shown of how accurately previous seasons were predicted. The analysis predicts that North Carolina will be the 2016 tournament winner.

1. © IBM 2015 1#WatsonAnalytics
Get started at
WatsonAnalytics.com
Ian Balina
WATSON ANALYTICS EVANGELIST
ibalina@us.ibm.com
Predicting March Madness
with IBM Watson Analytics

2. © IBM 2015 2#WatsonAnalytics
Agenda
• Intro to Watson Analytics
• Data Preparation
• Predictive Analysis
• Comparison of Predictive Model
• Predicting 2016 March Madness

3. © IBM 2015 #WatsonAnalytics 3
IBM Watson
To focus on bringing Watson to market, three
key areas were identified:
• Natural language processing
• Hypothesis generation
• Evidence-based learning
Trained using artificial intelligence (AI) and
machine learning algorithms to sense,
predict, infer, and in some ways, think.
The Watson you know that won Jeopardy! is now an IBM brand for a next generation of
cognitive computing solutions.

4. © IBM 2015 #WatsonAnalytics 4
How does Watson Analytics fit into the family of solutions?
Watson Analytics leverages Watson’s unique cognitive capabilities
in three ways:
1. Provides semantic recognition of concepts in your data such
as time, place, and revenue.
2. Identifies recommended starting points for analysis and
targets for prediction.
3. Enables interaction in natural language allowing for the
exploration of data through simple questions like you would
ask a colleague.
Watson Analytics is a cloud based smart data discovery tool from IBM Analytics

5. © IBM 2015 #WatsonAnalytics 5
Data Preparation
• Source - http://bit.ly/1nPJqma
• Seasons – 2001-2015 men’s college basketball seasons
The hardest part of this analysis is finding the data and merging data sets
• Combine all 15 seasons into one SUPER data set
• Add two columns for Final Four and Championship teams for each year

6. © IBM 2015 #WatsonAnalytics 6
Predictive Analysis
Top Predictors for Final Four teams
I uploaded the SUPER data set into Watson Analytics and set Final Four and
Championship as the targets to predict
Top Predictors for Championship Teams

7. © IBM 2015 #WatsonAnalytics 7
Historical Comparison of Predictive Model

8. © IBM 2015 #WatsonAnalytics 8
Predicting 2016 March Madness
North Carolina
Predicted
Winner

9. © IBM 2015 9#WatsonAnalytics
WatsonAnalytics.com – get started today!

10. © IBM 2015 10#WatsonAnalytics
Explore
Continue to ask
questions about
your data using
natural language
Click on the new
insights provided by
Watson Analytics and
explore the text and
visual results

11. © IBM 2015 11#WatsonAnalytics
Explore
Share images
with your peers,
or email it to
your team

12. © IBM 2015 12#WatsonAnalytics
Explore
Add in data points
and see things
differently using the
visualization menu,
data filters, color
pallets and more

13. © IBM 2015 13#WatsonAnalytics
Predict
Additional
insights
discovered
for you
Pin your results
for use in
dashboards and
storytelling
Automated data
visualizations
Identified
predictors of
your target
analysis

14. © IBM 2015 14#WatsonAnalytics
Predict
Additional tabs for
further predictive
insights
Pinning for
dashboard and
storytelling
Top decision rule
determined for
highest
percentage of
Wins

15. © IBM 2015 15#WatsonAnalytics
Assemble
Include graphics,
corporate logos
and more to
personalize your
view
Create
dashboards and
infographics
from your pinned
objects
Select from a
wide array of
templates and
color schemes