Dynamic network and data science applications in finance, security and genomics

Loading...
Thumbnail Image
Date
2024-08-21
Authors
Ranathungage, Thimani Dananjana
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Data Science studies are prolific in many application areas from health to finance and from supply chain to computer network security with varying objectives. This thesis is focusing on pattern mining and network analysis of data in some of these application areas. In the study of pattern mining, our focus is to compute waiting time in observing a desired pattern in three different application areas: patterns in DNA sequences, patterns in unauthorized access to computer systems and patterns of price rise and drops in stock prices. A novel fuzzy transition probability (TP) matrix is introduced, and a novel pattern mining algorithm is proposed for sequence data of any length. The proposed algorithm, which avoids the inversion of the pattern matrix, is applicable to Markov chains with huge state spaces. Furthermore, it is illustrated with real data that the incorporation of fuzzyness to the transition probability matrix is crucial in obtaining realistic forecasts. In the second study, we propose a novel method based on financial networks and their PageRank scores to form pairs to apply in pairs trading. In order to illustrate the practical performance of the proposed methods, algorithmic trading profits using the commonly used cointegration method are compared with the profits made by proposed correlation-based financial networks. The proposed method offers an advantage over the commonly used cointegration method by identifying more profitable trading pairs (stocks) for pairs trading.
Description
Keywords
Fuzzy stochastic matrices, Dynamic trading strategies
Citation