Improving the performance of machine learning algorithms through increasing dataset size

Abstract

Machine learning a very important field in computer science is utilized in many scientific domains and ever-widening range of human activities. Its main objective is to enable a machine to learn from past data, construct accurate predictive models and apply these models to a variety of problems such as classification. This ability has proven to be very effective in a variety of domains such as healthcare and business. One of the most important factors that determines if a Machine learning algorithm is successful in building a good predictive model or not, is the data available for analysis. Nowadays we are seeing a shift from having limited amount of available data to more data that we can store, analyse and process.

In this study, a set of experiments were designed and implemented to investigate the effect of increasing dataset size given to a Machine learning algorithm. Several datasets, Machine learning algorithms and evaluation techniques where made use of. The datasets used were split up into a number of increasing data size segments, each of which analysed and evaluated in terms of accuracy, cost and other perspectives. Each experiment yielded a range of results which led to a set of conclusions of interest. Whilst by increasing the dataset size the processing power needed to analyse this data also increases; it cannot be said that increasing the data size always resulted in a better performance. Another aspect was that other variations such as Machine learning algorithms and evaluation techniques had an important effect on the performance when increasing dataset size.