Learning Objectives:

  • Define machine learning and distinguish between supervised and unsupervised learning.

  • Explain key concepts: features, labels, training, testing, and model evaluation.

  • Understand the relationship between machine learning and AI.

2.1 Machine Learning and Its Role in Finance

Machine learning (ML) is a subset of artificial intelligence that enables systems to learn from data and improve over time without explicit programming. The NUS course introduces the concepts of data science, machine learning, and AI, along with detailed discussions on ML algorithms and specific use cases in finance . ETH Zürich’s course introduces different ML methods and discusses their application to problems in finance and insurance .

Machine learning is particularly valuable in finance because it can:

  • Process vast amounts of data faster than humans.

  • Identify complex patterns and relationships.

  • Adapt to changing market conditions.

  • Automate decision-making processes.

2.2 Core Machine Learning Concepts

The RPTPU course organises ML fundamentals into several key areas :

  • Types of Machine Learning:

    • Supervised Learning: Learning from labelled data to predict outcomes (e.g., credit scoring, fraud detection).

    • Unsupervised Learning: Finding patterns in unlabelled data (e.g., customer segmentation, anomaly detection).

    • Reinforcement Learning: Learning optimal actions through trial and error (e.g., algorithmic trading).

  • Key Terminology:

    • Features: Input variables used to make predictions.

    • Labels: The output variable being predicted.

    • Loss Function: A measure of how well the model performs.

    • Gradient Descent: An optimisation algorithm used to train models.

The University of Warsaw course covers these concepts in detail, including the types of predictions, types of models, and types of tabular data structures .

2.3 The Machine Learning Workflow

The typical ML workflow in banking follows a structured process:

  1. Problem Definition: Defining the business problem to be solved.

  2. Data Collection and Preparation: Gathering and cleaning relevant data.

  3. Feature Engineering: Selecting and transforming variables.

  4. Model Selection: Choosing appropriate algorithms.

  5. Training: Fitting the model to the data.

  6. Evaluation: Assessing model performance.

  7. Deployment: Integrating the model into business processes.

The University of Warsaw course includes a “Python – lightning fast course” to equip students with practical skills in data preparation, manipulation, and modelling using tools like NumPy, Pandas, Matplotlib, and Sklearn .

 

PreviousNext