Introduction to Data Science Challenges

Question 1

Finding data

Answer 1

• There may be hundreds or thousands of tables.

* There may be many different entities that are less relevant.

Question 2

Transforming data

Answer 2

• Reorganizing data, filtering, etc.

* Extracting relevant features.

Question 3

Dealing with data

Answer 3

Dealing with Big data

Dealing with streaming data

Question 4

Data quality

Answer 4

• Data may be incomplete, invalid, inconsistent, imprecise, and/or outdated.

Question 5

Fitting the data

Answer 5

Overfitting / underfitting

Question 6

Dealing with concept drift

Answer 6

Do Nothing (Static Model) The most common way is to not handle it at all and assume that the data does not change. …

Periodically Re-Fit. …

Periodically Update. …

Weight Data. …

Learn The Change. …

Detect and Choose Model. …

Data Preparation.

Question 7

Making results actionable

Answer 7

• Analysis results need to be relevant, specific, novel and clear.

Question 8

Ensuring fairness

Answer 8

Data science without prejudice: how to avoid

unfair conclusions even if they are true?

Question 9

Ensuring accuracy

Answer 9

Data science without
guesswork: how to answer
questions with a guaranteed level of accuracy?

Question 10

Ensuring confidentiality

Answer 10

Data science that
ensures confidentiality:
how to answer questions without revealing secrets?

Question 11

Ensuring transparency

Answer 11

Data Science that
provides transparency:
how to clarify answers such that they become
indisputable?

Question 12

Ill-posed problems

Answer 12

• A problem is well-posed if
− a solution exists and
− the solution is unique.

Question 13

• Problems in data science are often ill-posed

Answer 13

− there may be many possible models explaining observed phenomena

− the (training) data set is just a sample

− there may be noise (exceptional or incorrectly recorded instances) in the data set,

− the result needs to generalize to have predictive or explanatory value.