A joint analysis of heterogeneous information on Italian listed firms

Abstract

In recent years, the amount and variety of data available in digital format have exponentially increased. New technologies enables collecting, storing, transferring, and analysing huge amounts of data. Large digital archives, containing heterogeneous data (texts, figures, images, sounds) are easily available. The challenge is to extract useful information, by means of new methodological and computational tools, or with well-known tools used in innovative ways. Here we evaluate firms’ performances, jointly analysing financial measures and management commentaries. The data structure consists of two matrices, sharing the same rows (firms), a document-term matrix and a numerical matrix. In the framework of geometric data analysis, we use a graphical approach aiming at visualising both textual descriptions and financial indices.