Business Analytics - Case Study Quiz 1

Modules 1A and 1B

Data Mining (DM)

Process of analyzing large datasets to discover patterns and insights. Uses AI and statistics. Examples include fraud detection and customer behavior analysis.

Business Analytics (BA)

Transforms data into actionable insights that improve business decisions. Combines statistical methods and DM techniques.

Business Intelligence (BI)

Converts data into meaningful information for executives and managers.

Supervised Learning

Has DEPENDENT variable. S

Supervised Learning

Red for prediction and classification.

Supervised Learning

Examples: Regression, Logistic Regression, Decision Trees.

Unsupervised Learning

NO dependent variable.

Unsupervised Learning

Used to find hidden patterns and groups.

Unsupervised Learning

Examples: Clustering and Association Analysis

CRISP-DM

Cross Industry Standard Process for Data Mining

How many phases are there in CRISP-DM?

Six

What are the phases of CRISP-DM?

1. Business Understanding

2. Data Understanding

3. Data Preparation

4. Modeling

5. Evaluation

6. Deployment

Business Understanding

Define business objectives, assess the situation, convert business goals into technical DM goals, and create an initial hypothesis. Neglecting this phase can result in solving the wrong problem.

Data Understanding

Collect data, identify relevant variables, explore datasets, and understand available data sources. Sources include internal systems, external data providers, and surveys/research.

Data Preparation

Tasks include cleaning data, handling missing values, handling outlier, standardizing formats, removing redundancies, and transforming variables.

ETL Process

Extract, Transform, Load

Modeling

Choose models based on the problem type.

Continuous depending variable

Regression, Forecasting

Categorical dependent variable

Logistic Regression, Classification Trees

No dependent variable

Clustering, Association Models

Descriptive analysis type

WHAT happened?

Predictive analysis type

What WILL happen?

Prescriptive analysis type

What SHOULD happen?

Training set

Build the model

Validation set

Tune and refine the model

Test set

Measure final performanceO

Overfitting

Model performs extremely well on training data but poorly on new data.

Evaluation

Determines whether the model answers the business question and provides value.

Deployment

Includes implementing the model, reporting findings, monitoring performance, and maintaining systems.

Success Metric

Predictive accuracy, balanced with Cost-Benefit Analysis

BA turns data into ___

Business Insights

DM finds ___ in large data sets.

Patterns

Supervised Learning has a ___ variable.

Dependent

Unsupervised Learning has ___ dependent variable.

No

Regression predicts ___ outcomes.

Continuous

Logistic Regression predicts ___ outcomes

Categorical

Clustering is ___

Unsupervised

Association Analysis finds ___.

Product Relationships

CRISP-DM has ___ phases.

Six

Business Understanding created the ___.

Initial Hypothesis

ETL

Extract, Transform, Load

Training ___ the model

Builds

Validation ___ the model.

Fine-tunes

Test data measures ___.

Final Accuracy

Overfitting ___ performance on new data.

Hurts

Which learning method has a dependent variable?

Supervised Learning

What does ETL stand for?

Extract, Transform, Load

Which technique is used when no dependent variable exists?

Business Understanding

What is the most common DM success metric?

Predictive Accuracy