Programming language trends : an empirical study

Predicting the evolution of software engineering technology trends is a dubious proposition. The recent evolution of software technology is a prime example; it is fast paced and affected by many factors, which are themselves driven by a wide range of sources. This dissertation is part of a long term project intended to analyze software engineering technology trends and how they evolve. Basically, the following questions will be answered: How to watch, predict, adapt to, and affect software engineering trends? In this dissertation, one field of software engineering, programming languages, will be discussed. After reviewing the history of a group of programming languages, it shows that two kinds of factors, intrinsic factors and extrinsic factors, could affect the evolution of a programming language. Intrinsic factors are the factors that can be used to describe the general desigu criteria of programming languages. Extrinsic factors are the factors that are not directly related to the general attributes of programming languages, but still can affect their evolution. In order to describe the relationship of these factors and how they affect programming language trends, these factors need to be quantified. A score has been assigued to each factor for every programming language. By collecting historical data, a data warehouse has been established, which stores the value of each factor for every programming language. The programming language trends are described and evaluated by using these data. Empirical research attempts to capture observed behaviors by empirical laws. In this dissertation, statistical methods are used to describe historical programming language trends and predict the evolution of the future trends. Several statistics models are constructed to describe the relationships among these factors. Canonical correlation is used to do the factor analysis. Multivariate multiple regression method has been used to construct the statistics models for programming language trends. After statistics models are constructed to describe the historical programming language trends, they are extended to do tentative prediction for future trends. The models are validated by comparing the predictive data and the actual data