Introduction To Data Mining And Machine Learning

Introduction To Data Mining And Machine Learning

ADM 3308: Business Data Mining - Telfer School Of Management, University Of Ottawa

Table of Contents

• Introduction

• Machine Learning vs. Statistical Analysis

• Definition of Data Mining and Machine Learning

• Data Mining and Ethical Issues

• Data Mining Models

Introduction

• Overview of the course focused on Data Mining and Machine Learning concepts applicable in business contexts.

Machine Learning vs. Statistical Analysis

• Machine Learning and Statistical Analysis are often compared and contrasted.

  • Machine Learning: Focuses on developing algorithms that allow computers to learn from and make predictions based on data.

  • Statistical Analysis: Involves analyzing data to derive conclusions based on a set of assumptions, often employing hypothesis testing.

Definition of Data Mining and Machine Learning

• Data Mining: The exploration and analysis of a large quantity of data to discover meaningful patterns and rules.

• Knowledge Discovery in Data (KDD): The non-trivial process of identifying valid, novel, potentially useful, and understandable patterns in data.

  • Key Features of KDD:

  • Exploratory data analysis

  • Data-driven discovery

  • Deductive learning

Data Mining and Ethical Issues

• Discussion on ethical implications associated with data mining practices, including:

  • Privacy concerns surrounding user data.

  • Challenges of anonymizing data effectively.

  • Statistical data showing that 85% of Americans can be identified using just zip code, birth date, and gender.

  • Risk of discriminatory practices in data mining, such as in loan approvals, if using sensitive attributes like gender or race.

Data Mining Models

• Different data mining models and techniques include:

  • Classification: Grouping data into pre-defined classes (e.g., grades A, B, C, D).

  • Example: Approving loans based on attributes like age, region, and income.

  • Prediction/Estimation: Making predictions such as forecasting stock prices.

  • Clustering: Similar to classification but without pre-defined classes (e.g., clustering customers based on age or income).

  • Association Rules (Affinity Grouping): Identifying patterns like: if people buy product X, they often buy product Y with a given confidence percentage (Z%).

Data Structures

• Various sources and structures of data explored, including:

  • Business & Commerce: Corporate sales, stock transactions, etc.

  • Humanities and Social Sciences: Scanned books, historical documents, etc.

  • Entertainment: Internet images, streaming media.

  • Scientific Data: Data from astronomy, biology, etc.

  • Internet of Things (IoT): Data generated by machines and sensors.

What To Do With These Data?

• Addressing the question of deriving actionable insights from vast amounts of data generated across industries.

Data Mining Uncovers Hidden Information

• The process of data mining involves identifying and extracting hidden patterns and information within a dataset.

Multi-Disciplinary Field of Science and Technology

• Data Mining is recognized as a multi-disciplinary field, bridging various domains including computer science, statistics, and domain-specific areas.

Machine Learning

• Involves understanding both available data (training data) and new instances of data (unseen data).

• Applications of machine learning in business analytics include predictive use cases and improving data models.

Classification Example: Approving Loans

• Inputs for classification include attributes such as age, region, income, etc., leading to outcomes of either approved or not approved for loans.

Data Mining Models and Techniques

• Data mining includes various models focusing on:

  • Descriptive Models: Describing data characteristics.

  • Predictive Models: Making predictions based on historical data.

Statistical Concepts

• Definitions in statistics include:

  • Population: The complete set of items under consideration.

  • Sample: A subset of the population used for analysis.

  • Statistic: A summary measure that describes a characteristic of the sample data.

Null Hypothesis, P-Values, and Q-Values

• Null Hypothesis: States that no effect or difference exists.

• P-value: The probability of observing the data, given that the null hypothesis is true.

• Q-value: The confidence level associated with the p-value, computed as the reverse of the p-value.

Machine Learning vs. Statistics

• Shared Techniques: Both domains utilize similar algorithms and methodologies.

• Differences:

  • Machine Learning often handles larger datasets compared to traditional statistical analysis.

  • Machine Learning focuses on data behavior and prediction capabilities, whereas statistical analysis emphasizes hypothesis testing and model validity.

  • The dynamic nature of Machine Learning contrasts with the static assumptions in traditional statistics.

Data Mining and Knowledge Discovery

• Bridges statistical theory with performance improvement learning models.

• The full process of data mining includes:

  • Data cleaning

  • Learning

  • Validation

  • Integration

  • Visualization of results

CRISP-DM Data Mining Process Model

• Outline of the Data Mining process includes:

  • Business Understanding

  • Data Understanding

  • Data Preparation

  • Modeling

  • Evaluation

  • Deployment

Ethical Issues in Data Mining

• Critical considerations in data mining include questions related to the following:

  • What conclusions can be legitimately drawn from the data?

  • Are resources and results put to good use?

  • Is the data biased?

  • Is the model explainable?

• Emergence of Responsible AI and Responsible Data Science discussions.

References

• Linoff, G., & Berry, M. (2011). Data Mining Techniques for Marketing, Sales, and Customer Relationship Management, 3rd Edition. John Wiley.

• Shmueli, G., Bruce, P.C., Deokar, A.V., & Patel, N.R. (2023). Machine Learning for Business Analytics: Concepts, Techniques and Applications in RapidMiner. John Wiley.