E1

Question 1

Regression

Answer 1

Attempts to estimate or predict the numerical value of some variable for an individual (e.g. price of Microsoft stock).

Question 2

Classification

Answer 2

Attempts to predict which of a (small) set of classes an individual belong to.

Usually classes are mutually exclusive (e.g. customers will churn or will not churn).

Question 3

Regression Mathematical Formula / Rule-based Formula

Answer 3

Mathematical Formula

• Linear Regression
• Logistic Regression

Rule-based Formula
- Regression Trees

Question 4

Classification Mathematical Formula / Rule-based Formula

Answer 4

Mathematical Formula

• Logistic Regression
• Support Vector Machines

Rule-based Formula
- Classification Trees

Question 5

Stages of a predictive modeling process

Answer 5

1. Define target
2. Collect data
3. Build a model
4. Predict outcomes

Question 6

1. Define target

Answer 6

It has to be a quantifiable target.

a. E.g. what will be the stock price of Microsoft tomorrow? 200
b. E.g. will this client default on her loan? -> not specific enough! You want a time frame such as ‘first five years’ (though it depends on what you want to know)

Question 7

2. Collect data

Answer 7

We need data for the same or a related phenomenon

a. E.g. which customers defaulted last quarter? Think about whether or not this is the same phenomenon. Look at stuff that has happened in the past.

Question 8

3. Build a model

Answer 8

The model will look like a set of rules or a mathematical formula that allow establishing a prediction

a. E.g. rule: if (income <50k) then default, else no default.
b. E.g. mathematical formula: MSFTt+1 = 0.9 – APPLt -> predictive stock market value of Microsoft will be 90% of Apple’s.

Question 9

4. Predict outcomes

Answer 9

The model can be applied to any customer. It gives us a prediction of the target variable.

a. E.g. the customer will default because its income is lower than 50k.
b. E.g. MSFT = $212.