Python-la maşın öyrənməsiSyllabus

Subject code

İFS-31110yd

Credits

6

Student workload

180

Semester

Autumn semester

Academic year

2025-2026

Subject type

Elective

Form of education

Full-time

Language of instruction

Azerbaijani

In-class hours

a) 30 saat - mühazirə b) 30 saat - seminar

Out-of-class hours

a) 72 saat - sərbəst iş b) 48 saat - hazırlıq

“maşın öyrənməsi” fənni maşın öyrənməsinin nəzəri əsaslarını, alqoritmlərini və python mühitində tətbiqini öyrədir. fənn çərçivəsində numpy, pandas, scipy, matplotlib və seaborn kitabxanaları, verilənlərin statistik təhlili, ilkin emalı (data preprocessing) və vizuallaşdırılması, scikit-learn mühitində nəzarətli və nəzarətsiz öyrənmə alqoritmləri, o cümlədən decision tree, random forest, knn, svm, naive bayes, reqressiya, k-means və DBSCAN metodları öyrənilir. kursun sonunda tələbələr maşın öyrənməsi modellərini qurmaq, qiymətləndirmək və real verilənlər üzərində tətbiq etmək, həmçinin generativ süni intellekt alətlərindən istifadə etməklə analitik problemlərin həllinə praktik yanaşmalar formalaşdırırlar.

• seminar
• presentations
• interactive discussion
• lecture
• project-based learning

1. 1

   CLO 1: Explains the basic concepts, types, development stages and application areas of machine learning.

2. 2

   CLO 2: Processes, analyzes and visualizes data using numpy, pandas, scipy, matplotlib and seaborn libraries.

3. 3

   CLO 3: Prepares data for machine learning by applying data preprocessing and statistical analysis methods.

4. 4

   CLO 4: Solves prediction and classification problems by applying supervised learning algorithms (decision tree, random forest, knn, svm, naive bayes and regression).

5. 5

   CLO 5: Performs data clustering by applying unsupervised learning algorithms (k-means, dbscan) and evaluates the results.

6. 6

   CLO 6: Builds, evaluates, and evaluates machine learning models and interprets results using generative artificial intelligence tools and applies them to solving practical problems.

1. 1.
   A numpy moduleLecture

   The topic explores the numpy library, the cornerstone of machine learning and high-performance scientific computing in python. students analyze multidimensional array structures that take up less memory than standard python lists and perform vectorized calculations on the fly. in the course of the lesson, the creation of arrays of different sizes, indexing, cutting, application of mathematical and statistical operators, as well as the translation mechanism that ensures that arrays of different sizes enter into mathematical operations with each other are investigated.

   • Processes, analyzes and visualizes data using numpy, pandas, scipy, matplotlib and seaborn libraries.
2. 2.
   Pandas moduleLecture

   In the topic, the pandas library is studied, which is the most widely used in data manipulation, cleaning and structuring for machine learning models in the python ecosystem. students analyze the concepts of the one-dimensional series and dataframe, the two-dimensional table structure that forms the core of the library. in the course of the lesson, the principles of reading data from files of different formats, selecting rows and columns, filtering according to conditions, managing missing data, and performing aggregate calculations by grouping data are examined.

   • Processes, analyzes and visualizes data using numpy, pandas, scipy, matplotlib and seaborn libraries.
3. 3.
   Statistical analysis of dataLecture

   The subject studies statistical analysis methods applied to understand the internal structure, distribution and characteristics of data before building machine learning models. students analyze methods of descriptive statistics, measures of central tendency, and degrees of dispersion of variables that reveal basic characteristics of a data set. in the main part of the lesson, the calculation and interpretation of correlation matrices defining variance, standard deviation, quartiles, as well as dependencies of raw data are practically explored through python's pandas, numpy and statistical modules.

   • Prepares data for machine learning by applying data preprocessing and statistical analysis methods.
4. 4.
   Giriş. dövlətəqədərki dövr və ibtidai insan münasibətlərinin formalaşması.Seminar

   Dövlətəqədərki dövrün əsas xüsusiyyətləri, ibtidai insan cəmiyyətinin yaranması və formalaşması, eyni zamanda qədim insanların həyat tərzi, əmək vərdişləri, ictimai münasibətlərin təşəkkülü, qəbilə və tayfa quruluşunun formalaşması, eləcə də dövlətin yaranmasına təsir edən sosial-iqtisadi proseslərdən bəhs edir.

   • • ftn2: azərbaycan tarixi üzrə bəzi mübahisəli məsələlərə aydınlıq gətirilir.
   • • ftn1: azərbaycan tarixinin dövrlərinin əsas mahiyyəti izah edilir.
   • • ftn3: tələbələrin bu və ya digər məsələlərlə bağlı obyektiv mövqeyini formalaşmasına nail olunur.
   • • ftn4: tələbələrin ən azından minimum məlumat əldə etməsi, mövzular ətrafında öz fikrinin formalaşmasının səviyyəsi yoxlanılır.
   • • ftn5: tələbələrə qədim tariximizin müasir dövrlə uzlaşdırılması vərdişləri aşılanır.
5. 5.
   Data visualization: seaborn and matplotlib modulesLecture

   The topic explores two key libraries — matplotlib and seaborn — used for visual data analysis, graphing, and effective data storytelling in the python environment. students analyze matplotlib's flexible low-level structure, customization of plot elements, and high-level statistical visualization capabilities that integrate directly with seaborn's pandas dataframe structure. in the course of the lesson, the construction of line graphs showing trends, histograms showing distribution, heat maps showing correlation, and diagrams defining scatter plots are practically explored.

   • Processes, analyzes and visualizes data using numpy, pandas, scipy, matplotlib and seaborn libraries.
6. 6.
   Data visualization: matplotlib moduleLecture

   In the topic, the matplotlib library, which is considered the cornerstone for visual data analysis and graphing in the python ecosystem, is studied. students analyze the library's low-level flexible architecture, object-oriented graphics structure, and customization capabilities of graphics elements. programming of line graphs for tracking continuous trends, bar charts for cross-category comparisons, scatter plots for viewing data dispersion and correlation, and histograms for analyzing distributions is explored during the lesson.

7. 7.
   Data preprocessing with scikit learnLecture

   The subject studies the preprocessing methods provided by the scikit-learn library to bring raw data into a clean and optimal format that can be accepted by machine learning algorithms. students analyze min-max scaling and standardization techniques to remove the negative effect of different scaled features on the model. in the course of the lesson, the labelencoder and onehotencoder tools for converting categorical variables into numerical format, as well as the practical application of the simpleimputer class for filling gaps in data, are explored.

   • Prepares data for machine learning by applying data preprocessing and statistical analysis methods.
8. 8.
   K-nearest neighbor (knn) algorithmLecture

   The k-nearest neighbor algorithm, which is one of the simplest, instance-based and lazy learning algorithms of supervised learning, is studied in the subject. students analyze that the model does not construct any mathematical equations during the training phase, but simply stores the data in memory and evaluates the newly arrived data point based on the category of its "k" number of neighbors that are closest to it in terms of position. the course examines the euclidean, manhattan, and minkowski metrics used to measure the distance between points, as well as the principles of choosing the optimal "k" parameter, which directly affects the performance of the model.

   • Solves prediction and classification problems by applying supervised learning algorithms (decision tree, random forest, knn, svm, naive bayes and regression).
9. 9.
   Support vector machine (svm) and naive bayes algorithmsLecture

   Two powerful and different classification algorithms of supervised learning model based on mathematical approach — support vector machine and naive bayes — are studied in the topic. students analyze the working principle of the SVM algorithm, which finds the most correct boundary that optimally separates a group of data, as well as the kernel trick, which upscales non-linear data. in the second part of the lesson, the logic of the naive bayes algorithm, which is based on probability theory and bayes' theorem, assumes that the features are independent of each other, and works at high speed in text classification tasks, is investigated.

   • Solves prediction and classification problems by applying supervised learning algorithms (decision tree, random forest, knn, svm, naive bayes and regression).
10. 10.
    Regression analysis and its use in machine learning.Lecture

    Regression analysis, which is one of the fundamental pillars of both classical statistics and supervised machine learning, is studied in the subject. students analyze the mechanism of establishing mathematical relationships between a dependent variable and one or more independent variables. the course examines the differences between linear and non-linear regression models, minimizing the sum of squared errors, metrics such as mse, rmse, and MAE used to evaluate model performance, as well as industrial applications such as price forecasting, trend analysis, and risk assessment.

    • Solves prediction and classification problems by applying supervised learning algorithms (decision tree, random forest, knn, svm, naive bayes and regression).
11. 11.
    Introduction to unsupervised learning. k-means algorithmLecture

    The subject is studying the field of unsupervised learning, where machine learning has no output labels but detects hidden patterns in data based only on input features. students analyze the working principle of k-means, the most popular and fundamental clustering algorithm used to discover the internal structures of data. in the course of the lesson, the random determination of center points, grouping of data according to distance metrics, the mechanism of recalculating the centers within the operation and the elbow method used for finding the optimal number of clusters are investigated.

    • Performs data clustering by applying unsupervised learning algorithms (k-means, dbscan) and evaluates the results.
12. 12.
    Dbscan clustering algorithmLecture

    In the topic, the DBSCAN algorithm is studied, which is based on density based unsupervised learning and is used to detect clusters of data with complex geometric shape. students analyze the characteristics of dbscan, which, unlike the center-based k-means algorithm, does not require a priori cluster count and can successfully identify outliers. in the course of the lesson, the basic parameters of the algorithm, the radius and the minimum number of neighboring points, as well as the mechanism of classifying data into core, boundary and noise are practically investigated.

    • Performs data clustering by applying unsupervised learning algorithms (k-means, dbscan) and evaluates the results.
13. 13.
    Generative artificial intelligence: its use in building machine learning models and interpreting results.Lecture

    The subject studies methods of applying big language models and generative artificial intelligence tools to accelerate and automate machine learning processes. students analyze the role of generative tools in generating synthetic data for balancing datasets, code optimization in the feature engineering phase, and building model architectures. the main part of the lesson explores the capabilities of generative intelligence in interpreting the decisions, error matrices and performance metrics made by complex machine learning models into business language.

    • Builds, evaluates, and evaluates machine learning models and interprets results using generative artificial intelligence tools and applies them to solving practical problems.

1. Andreas c. müller, sarah guido - "introduction to machine learning with python" (2016)(opens in a new tab)
2. Mark e. fenner - "machine learning with python for everyone" (2019)(opens in a new tab)
3. Andreas c. müller, sarah guido - "introduction to machine learning with python"(opens in a new tab)