Machine Learning in Finance I 2400-QFU1MLF

The course is conducted in two sequential stages: 1) lectures 2) case study laboratories. The lecture stage is followed by an exam which verifies the required theoretical knowledge. On the other hand, the laboratory stage is followed by practical projects realised by the student.

I part - Lectures

1. Introduction to Machine Learning
a. Types of Machine Learning
b. Introduction to Supervised Statistical Learning
c. Types of predictions, types of models, types of tabular data structures
d. Notations and general concepts - loss function, cost function, gradient descent
e. Simple Supervised Learning models - linear regression and logistic regression

2. Assessing model accuracy, machine learning diagnostics
a. Evaluation metrics- regression and classification
b. Learning curves
c. Training, validation and testing sets
d. Cross-validation technique
e. The concept of bias and variance and their trade-off
f. Possible remedies of underfitting or overfitting

3. Basic Supervised Learning models
a. K-nearest neighbours
b. Support Vector Machines
c. Decision trees and Random Forest (bagging idea)

4. Crucial machine learning techniques
a. Dataset preparation steps
b. Initial feature selection methods
c. Feature engineering
d. Regularization
e. Rebalancing
f. Explainable Artificial Intelligence*

Exam, in the convention of recruitment questions for the position of Data Scientist, Quantitative analysis, Machine Learning Engineer

II part - Labs

5. Python - lightning fast course
a. Preparation of the environment
b. Variables, data type, operators and control structure
c. Functions
d. Modules
e. Data science toolkit: NumPy, Pandas, Matplotlib, Sklearn

6. Case study - credit risk modeling
a. Construction of the first classification machine learning end-to-end pipeline
b. Solving the problem of unbalanced data set
c. Testing multiple machine learning models
d. Comparing the results of the models in the business context
e. Solving the problem of explainability of the solution*

7. Case study - medical insurance premium prediction
a. Construction of the first regression machine learning end-to-end pipeline
b. Testing multiple machine learning models
c. Comparing the results of the models in the business context
d. Solving the problem of explainability of the solution*

8. Case study - simple algorithmic trading
a. Construction of the first time series machine learning pipeline
b. Solving the problem of specifics of data preparation for the problem of time series
c. Testing multiple machine learning models
d. Playing a simple investment strategy based on trained model

Additional case study - life insurance assessment (multiclass classification)

9. Project presentations