PrefStat organizes a series of summer schools designed to provide a comprehensive overview of preference statistics, a rapidly growing field that has gained significant attention in recent years due to the numerous application fields involving human preferences (from recommender systems to Large Language Models, from surveys and psychological experiments to marketing, economics, and political science). PrefStat thus establishes a series of high-level courses on cutting-edge topics in the specific context of Statistical Learning from Preference Information, or Preference Learning. Preference learning is concerned with all data analyses involving preferences, rankings, ratings, clicking, or any kind of ordinal data. It entails modeling experiments involving a set of assessors (experts, judges, users) who express order relations about a set of items, thus being a subfield of both supervised and unsupervised statistical learning. The school will provide a deep introduction to the topic and insight into more challenging tasks that are of interest in modern applications, such as handling partial, unstructured, exogenous information, individual preference prediction, and importance feature selection. PrefStat 2025 will combine lectures delivered by internationally leading scholars on the specific designated topic and supervised practical tutorials.