Mathematics for Data Science. Part 3. Optimization methods and data analysis algorithms - course RUB 32,490. from Specialist, training 40 academic hours, date of May 15, 2023.
Miscellaneous / / December 03, 2023
Professional programming course teacher, certified developer Python Institute with general work experience in the IT field more than 20 years. Built IT systems in 4 companies from scratch. More than 5 years.
Vadim Viktorovich graduated from the Russian State University for the Humanities in 2000 with a specialization in Informatics and Computer Science. A true professional in administration matters DBMS, automation of company business processes (ERP, CRM etc.), creating test cases and training employees.
Able to motivate and captivate. He is demanding of his listeners, always ready to clarify difficult points. Extensive experience working on real projects allows him to pay attention to those details that are usually overlooked by novice developers.
Module 1. Optimization methods (16 ac. h.)
- Basic concepts, definitions, subject
- Continuity, smoothness and convergence of digital functions. Discrete digital functions
- Conditional and unconditional optimization
- Single-criteria optimization methods
- Statement of the multicriteria optimization problem
- Multicriteria optimization methods
- Gradient descent
- Stochastic optimization methods
Module 2. Data analysis algorithms (16 ac. h.)
- Linear regression algorithm. Gradient Descent
- Scaling of features. L1- and L2-regularization. Stochastic gradient descent
- Logistic regression
- Algorithm for constructing a decision tree. Random forest
- Gradient boosting
- Analysis of the backpropagation algorithm
Module 3. Final work (8 ac. h.)
Data science includes a wide range of approaches and methods for collecting, processing, analyzing and visualizing data sets of any size. A separate practically important area of this science is working with big data using new principles mathematical and computational modeling, when classical methods stop working due to their impossibility scaling. This course is designed to help the student learn the basics of the subject area through formulation and solving typical problems that a data science researcher may encounter in his or her work. To teach the student to solve such problems, the authors of the course provide the student with the necessary theoretical minimum and show how to use the tool base in practice.
4,2