Python training in the field of data science

The basic training in Python with the focus on data science is the ideal introduction for analysts to the programming language. In addition to programming paradigms, object types and syntax structures and polars, the central library for data management. Through the training the participants will be able to implement the data management as well as the first analyses and visualizations in Python.

Course content:

• First steps into Python
• Concept and philosophy of Python
• Data structures and their properties
• Importing data
• Data management with polars
• Data analysis with Python
• Loops and control elements

—

Recommended course length: Two days

A selection of the training content in the area of machine learning with Python:

• Introduction to the basic concepts of machine learning
• Using the machine learning framework scikit-learn
• Logic and practice: splitting data into test and training data
• Presentation and application of relevant processing steps, e.g. one-hot encoding, standardization, imputation
• Training of machine learning models based on various algorithms, e.g. decision trees, support vector machines, random forests, XGBoost, etc.
• Linking of preparation and modeling steps in a scikit-learn pipeline
• Interpretation of various metrics for model evaluation
• For classifications: (Balanced) Accuracy, AUC, F1-Score, etc.
• For regressions: RMSE, MAE
• Parameter tuning (grid search) of the models with the help of cross-validation
• Independent application of what has been learned in practical exercises

  -

  Target group: This introductory course is aimed at participants with previous knowledge of Python or statistics.

  -

  Course length: Two days

Transformer models have revolutionized the way computers process and understand text. They form the foundation of modern AI applications. In our training course, we look at the background and use cases of transformer models and teach the skills for their practical application.

The course content at a glance:

• Introduction: Presentation of the use cases of Transformer models (text, audio, image)
• Background of Transformer models
• Practical work with Transformers in Huggingface
• What is Huggingface and what do I need it for?
• How do I process data and prepare it for working with transformers? (datasets library)
• How do I apply Transformer models in different use cases? (transformers library)
• How do I evaluate the performance of the models? (evaluate library)
• How can I fine-tune a pre-trained model for my use case?

-

Course length: Two days

Optionally, we can extend our training on the topic of transformer models by a further day to a total length of three days.

The course content of the extension day at a glance:

• Transformers in production: Presentation of a RAG system
• Improve speed or reduce RAM consumption with quantization and the accelerate library
• What role do transformers play in a RAG system
• How do I set up a RAG system: langchain library
• Prompt Engineering
  (display RAG as an app with gradio)