This document describes a student's diploma thesis project on automatically estimating the importance of software tasks from project management data. It discusses using information from code repositories and issue tracking systems to select important characteristics and apply machine learning algorithms to predict task importance values. The goal is to automate and improve upon the current manual process of assigning importance, which can be time-consuming and inconsistent for large software projects. The document outlines the methodology, which involves preprocessing the data, creating text and numeric feature models, and evaluating different classification algorithms to select the best for this task.