Course Description

The course introduces participants to mathematical, primarily statistical, models as a tool for analysing social phenomena. The main focus is on interpreting a model as a "black box": understanding its purpose, its data requirements and the correct reading of results, without deriving formulas or mandatory programming. It covers the basics of statistics, supervised and unsupervised models with their parameters and quality metrics; a separate block addresses models for textual data, as well as causality and typical interpretation errors. The course is intended for participants from various academic fields and areas of activity. An optional lab for those proficient in programming and a block on building simple models independently are also provided.

Prerequisites

No special mathematical background is required: school-level mathematics and a willingness to work with data and tables are sufficient. Programming is not required — coding skills are needed only for the optional lab

Learning Goals
- forming an understanding of basic mathematical and statistical methods and models and their fields of application;
- developing skills in understanding and correctly interpreting the results of mathematical models and methods;
- mastering techniques for building basic mathematical models for the analysis of social phenomena;
- forming a critical attitude towards model and other methods conclusions: an awareness of their limitations, pitfalls and questions of causality.

Learning outcomes

Upon completion of the course the participant:
- understands the definitions of basic mathematical and statistical models and methods and their fields of application;
- is able to read and interpret model results correctly, including quality metrics;
- is able to apply basic methods and build simple models for the analysis of social phenomena;
- is aware of model limitations and typical interpretation errors, including those related to causality.

Class topics

1. The purpose of models and a typology of data. Introduction to modelling.

2. Basics of statistics: p-value, confidence intervals, the Bayesian approach.

3. Supervised and unsupervised models: linear, logistic and Bayesian regression; metrics R²/MAE/RMSE and precision/recall/F1.

4. Models and metrics for textual data: TF-IDF, embeddings, topic modelling, inter-annotator agreement.

5. Causality and typical pitfalls; foundations of building models independently.

Assessment

A simple model on a topic chosen by the participant, completed with the support of an IT mentor and defended publicly. The work includes:
- a 5–7 page research description (goal and research questions, methodology, data collection, analysis, conclusions);
- reproducible code;
- a presentation.

The emphasis is placed on correct framing of the problem and honesty of conclusions rather than on the complexity of the model.

The final grade consists of three mandatory components: homework - 40%, exam - 30%, ICR defense - 30%; For each component, it is necessary to score at least 50%. Optional laboratory is not included in the final score and gives up to +15% bonus

Examples of thesis topics

1. Factors affecting the life cycle of small businesses in Batumi;

2. Dynamics of one-bedroom apartment prices in Batumi over the past five years.