Pie & AI Kyiv: From Data to Prediction

Learn how raw data becomes a machine learning prediction in this practical, beginner-friendly online Pie & AI session from the Kyiv community.Hosted by the Pie & AI by DeepLearning.AI community, From Data to Prediction takes place online on 16 September 2026, from 16:00 to 17:15. The 75-minute session will be delivered in Ukrainian and is open to learners of all levels, with the material specifically designed to be accessible to beginners.## What you’ll learnThe meetup follows a complete machine learning workflow rather than discussing AI only in abstract terms. You’ll see how to move from data preparation to a trained model, then examine what its output actually means. Topics include:- How machine learning changes the traditional relationship between data, rules and outcomes- A beginner-friendly explanation of how neural networks learn patterns- Preparing financial market data and selecting useful features- Defining inputs, targets, training data and test data- Building, training and evaluating a simple neural network with Python and TensorFlow/Keras- Visualising predictions and assessing model performance- Using SQL and Google Cloud BigQuery ML as an alternative route to predictive analytics- Understanding data quality, overfitting, model limitations and responsible interpretationFinancial market data provides the educational case study, helping make the concepts concrete without presenting the model as investment advice or claiming that future market prices can be predicted reliably.## Session formatThe event combines concise explanations with a practical demonstration. After a welcome and introduction, the session progresses from traditional programming to machine learning, covers neural network fundamentals, and turns financial data into a defined ML problem. The central demo shows a TensorFlow/Keras model being built, trained and evaluated, followed by a shorter BigQuery ML comparison.The final section asks an essential question: what does a prediction actually tell us? An open discussion and Q&A will give attendees the opportunity to explore model performance, data limitations and practical applications.## Speaker and community…Register for this event