2. Agenda
Definition
What Impact?
Big Data into the Big Picture
Detecting early signals
Prevention
Key takeaways
3. Biography
Senior Project Manager at Ministry of Education
Certified Professional in BCM
Data architect, Big Data Ops
25+ years in IT Services
Data Natives Ambassador
“I’m a positive individual, yet skeptical”
3
4. Definition
Leaving high school, college, university or
another group for practical reasons, necessities,
or disillusionment with the system from which
the individual in question leaves.
Withdrawal from established society,
especially to pursue an alternate lifestyle.
5. Common Dropout Characteristics
• Demographic factors
– Socio-economic characteristics of a population expressed
statistically: age, sex, education level, income level,
marital status
• Health issues
– Chronic diseases
Asthma, Diabetes, Cancer, AIDS, Epilepsy, Congenital heart
issues,..
– Mental Illness
Disorders: Anxiety, Bipolar, Depression, Obsessive-
Compulsive, Schizophrenia,..
6. Different risks affecting individuals
Parents
Physical
Environment
Personality
“Humans are physical, biological, psychological, cultural, social, historical beings. This complex unity of
human nature has been so thoroughly disintegrated by education divided into disciplines, that we can
no longer learn what human being means.”
Pascal Morin
7. Consequences of Dropout
Decreases the talent pool of a Nation
Less earnings and less income to the Economy
Influx of government expenses (Education,
Justice, Healthcare)
More violence and loneliness
Less engagement among teenagers
8. Integrate IT: how frequently is the
student logging-in into his/her
account?
Social Data: time spent on Social
Media on a daily basis
Clickstream data and sentiment
analysis
Detection by keywords on the web
(Google search, blogs, Tweets,
Instagram, FB groups..)
Big Data into the Big Picture:
Detecting early Signals
9. • Social assistance
• Special programs (awareness)
• Regular Follow-up
• Sports
Big Data into the Big Picture:
Preventing Dropout
10. 10
Key Takeaways
• What worked well
People: Team building
Tools, Data sources
Data availability, its capture, auditing and Master
data management
• Improvements to be expected
MAD skills
Legal & regulatory requirements (access, analyze,
share)
Supervised methods and set of training data: how
big is enough?
Discrete outcomes (Y/N), and thresholds to be set (a
probability being returned with logistic regression
approach <> binary classification problems)
Test, test, test..