Spreadsheets compose a notably large and valuable dataset of documents within the enterprise settings and on the Web. However, extracting data from these files it is rather a cumbersome task. They are optimized for human consumption, but lack with what regards automatic machine processing. In this thesis, we work on the cell level to infer their layout function within the sheet. Our immediate goal is to form an initial understanding (basis) the contents. In the future, starting from this initial basis we aim at identifying the tables in the sheet, and subsequently extract their schema and data. Related work has already proposed different cell layout functions and techniques to infer them. We utilize, in this thesis, supervised machine learning. A considerable number of the cell features are defined by us, and the rest are reused from the literature. Moreover, we experimented with various sets (combinations) of layout functions (classes), in order to determine which one perform best. Our evaluation shows that some of the layout classes are obsolete. Once they are absorbed by more performant classes, the overall accuracy of the approach increases.