Data Informed Decision Bias
Data science techniques are revolutionizing decision making processes and facilitating data driven insights. The exponential growth of data availability, coupled with advancements in computing power and algorithms, has paved the way for a data driven paradigm that is reshaping the way organizations operate. In the present thesis we discuss the use of data science techniques for decision making. We first conduct a case study of using data science techniques to reveal latent drivers for improving societal outcomes. Secondly, we reveal class imbalance issues in datasets exploited for decision-making purposes. Furthermore, we present a comprehensive discourse on discriminatory bias within the framework of machine learning algorithms. For mitigating machine learning bias, we subsequently produce novel results at the intersection of Learning Fair Representations and Variational Autoencoders. We develop a novel approach in the field of fair representation learning that demonstrates comparable or superior performance when compared to existing state-of-the-art algorithms in the domain of representation learning.