Introduction to Data Science and Artificial Intelligence with Python

Target Audience

The course is designed for individuals who want to develop data/AI skills in practice, particularly:

• data analysts, business analysts, and reporting specialists
• professionals from finance, sales, operations, and marketing departments working with data
• developers and engineers who wish to solidify their foundations in ML/LLM using Python
• individuals preparing for a Data Scientist / ML Engineer role at a junior to regular level

Prerequisites

• basic knowledge of Python (variables, loops, functions, working in a notebook)
• basics of data handling (tables, data types, simple calculations)
• readiness to work on your own computer in a remote environment

Training Methods

Practical workshops dominate (approx. 75% of the time). During the sessions, we use:

• individual and team exercises in notebooks (Python)
• short theoretical introductions preceding labs
• case studies based on data resembling real-world business applications
• mini-projects summarizing each day and work with pipelines
• Q&A sessions and consultations regarding participants' work issues (if data/cases are provided)

Training Materials

Participants receive a complete set of materials used during the training, including:

• Jupyter notebooks and exercise files
• dataset packs for labs
• cheat sheets with key Pandas functions and scikit-learn elements
• links to recommended resources and documentation

Technical Conditions for Remote Training

Participation requires:

• computer with Windows/macOS/Linux OS (min. 8 GB RAM, 16 GB recommended)
• stable internet connection (min. 10 Mb/s)
• headset with microphone and camera (recommended for workshop work)
• access to MS Teams or Zoom communicator
• optional: access to the DaDesktop environment provided by NobleProg (if launched for the group)

Validation and Certificates

Validation of learning outcomes is based on practical tasks (notebooks), mini-projects, and quality checklists. Upon completion of the training, participants receive a NobleProg certificate in electronic format.

Information Document

The program can be tailored to the group's needs after an requirements analysis (PCQ).

Data Analysis with Pandas

• Creates and modifies Series/DataFrame objects, performs operations on tables and columns
• Applies filtering, sorting, grouping, and aggregations to solve analytical problems

Verification criteria: Verification: practical tasks in a notebook (calculation results and correctness of data transformations)

Regression Models

• Selects and trains a regression model for a given problem
• Evaluates the model using appropriate metrics and interprets results
• Applies normalization and data preparation, as well as prevents overfitting

Verification criteria: Verification: regression mini-project + discussion of metrics and insights.

Classification Models

• Builds and compares classification models, selecting metrics (e.g., accuracy, precision/recall, F1, ROC-AUC)
• Applies ensemble techniques to improve prediction quality

Verification criteria: Verification: classification task + comparison of at least 2 models with justification for the choice.

Working with LLMs

• Understands key concepts: prompts, context, embeddings, tokenization, model limitations
• Uses LLM APIs for text generation and processing and creates embeddings for semantic search
• Knows basic security and quality principles (data protection, testing, limitations)

Verification criteria: Verification: integration exercise (simple prototype) + data security checklist.

Day 1 - Data Analysis with Pandas

• Basic data types: Series and DataFrame (creation, indexing, data types).
• Table operations: loading data, merging/joining, concatenation.
• Filtering, sorting, grouping: groupby, aggregations, pivot tables.
• Value modification: mapping, replace, missing data (NaN) and imputation strategies.
• Column operations: feature creation and transformation, apply/assign functions.

Day 2 - Machine Learning: Regression Algorithms

• Introduction to regression: problem, data, ML pipeline.
• Model evaluation: metrics (MAE, MSE/RMSE, R2), train/test split.
• Data normalization and standardization: when and why.
• Handling categorical variables: One-Hot Encoding and Label Encoding.
• Overfitting: diagnosis and reduction methods.
• Cross-validation: selection of validation strategy.
• Grid Search and hyperparameter optimization (including pipeline).

Day 3 - Machine Learning: Classification Algorithms

• Introduction to classification: binary and multi-class.
• Classification evaluation: confusion matrix, accuracy, precision/recall, F1, ROC-AUC.
• Overview of classification algorithms (e.g., logistic regression, trees, SVM, kNN).
• Ensemble: combining classifiers (bagging, boosting, stacking - overview and practice).

Day 4 - Large Language Models (LLM)

• Introduction to LLMs: capabilities, limitations, typical business use cases.
• OpenAI API and other models: integration basics, costs, limits, best practices.
• Multimodal models: working with text and images - scenarios and limitations.
• Embeddings: semantic search, clustering, simple recommendations, RAG basics.