Магистратура
2024/2025
Современные методы принятия решений: прикладные задачи машинного обучения
Статус:
Курс обязательный
Направление:
01.04.02. Прикладная математика и информатика
Где читается:
Факультет компьютерных наук
Когда читается:
2-й курс, 2 модуль
Формат изучения:
с онлайн-курсом
Онлайн-часы:
60
Охват аудитории:
для своего кампуса
Прогр. обучения:
Магистр по наукам о данных (о)
Язык:
английский
Кредиты:
3
Контактные часы:
12
Course Syllabus
Abstract
Machine learning is a very popular and important field in the modern world. Every day lots of people deal with products that have been created using various machine learning technologies. In order to enable the machines to make right decisions based on data, different problems need to be addressed. The course is focused on the advanced tasks and instruments of data analysis and machine learning. The course is aimed at the participants who liked the basics of data analysis and machine learning and who want to study data science in more detail.
Practice parts are conducted in programming language Python and are based on different libraries such as numpy, pandas, matplotlib, scikit-learn, and others. In order to successfully pass the course, listeners are required to have mathematical skills at the school level, skills of programming in Python, and also a basic knowledge of data analysis and machine learning. Listener's performance is evaluated using programming assignments, theoretical tests based on the materials from lectures, and project assignment.
Learning Objectives
- After the end of the course, the listeners will: - learn how to process categorical data; - examine the details of different boosting methods and learn how to construct multi-level models using such ensembling techniques as blending and stacking; - know how to validate and interpret machine learning models; - be able to solve tasks with highly imbalanced datasets; - master different techniques of clustering, dimensionality reduction and data visualization; - gain skills in working with recommender systems.
Expected Learning Outcomes
- Learn how to process categorical data.
- Examine the details of different boosting methods and learn how to construct multi-level models using such ensembling techniques as blending and stacking.
- Know how to validate and interpret machine learning models.
- Be able to solve tasks with highly imbalanced datasets.
- Master different techniques of clustering, dimensionality reduction and data visualization.
- Gain skills in working with recommender systems.
Course Contents
- 1. Handling categorical data
- 2. Advanced ensembling techniques
- 3. Model verification
- 4. Handling imbalanced data
- 5. Recommender Systems
- 6. Clustering & Visualization
Bibliography
Recommended Core Bibliography
- Molnar, C. (2018). iml: An R package for Interpretable Machine Learning. https://doi.org/10.5281/zenodo.1299058
Recommended Additional Bibliography
- Machine learning : beginner's guide to machine learning, data mining, big data, artificial intelligence and neural networks, Trinity, L., 2019
- Machine learning in action, Harrington, P., 2012