The Need for Diagnostics for Classification Algorithms

Many machine learning researchers view the task of inductive generalization as beginning after the data is collected, assuming that the useful features have been identified and that representative data has been collected. This assumption has led researchers to focus, with considerable success, on algorithm development. As a result, little attention has been paid to applying machine learning algorithms. One problem that arises is that when classification performance does not meet expectations, inexperienced practitioners can find little guidance in the available literature to help them. This talk addresses this gap between research and applied machine learning and suggests areas of research that can help bridge this gap. 1 THE APPLICATION DEVELOPMENT PROCESS The first step in the application development process is to analyze the factors relevant to the application domain. Application factors include the overall objectives of the project, the amount of domain knowledge available, and th..