Machine Learning Algorithms for Wealth Data Analytics
- Publication Type:
- Thesis
- Issue Date:
- 2020
Open Access
Copyright Clearance Process
- Recently Added
- In Progress
- Open Access
This item is open access.
The thesis investigates multiple machine learning algorithms with big data approach and applies cutting-edge deep analytics to tackle the challenges in financial wealth management. In general, the existing research on wealth data analytics is limited with two main challenges. Firstly, the amount of quantitative research conducted is scarce and scattered across different approaches. Partially this is due to the lack of access to the data required for the research to use a quantitative approach. Secondly, the results are rudimentary and limited to a certain aspect of wealth data analytics. This lack of integration in existing research findings is a by-product of the simplistic approaches employed in lieu of big data analytics and deep learning techniques.
This research provides a broader and comprehensive approach for quantitative research within the wealth management field from both financial and customer aspects. Particularly, this research utilizes the big data of structured demographic, behavioral, communicational data, and unstructured textual information from wealth customers, plus additional financial market and corporate responsibility data from companies. This thesis exploits deep analytics techniques to provide a better framework for decision-making support based on the constructed mathematical and computational models, combined with customer segmentation modeling and quantitative finance approach.
From the customer aspect, the thesis applies big data analytics, text mining and interpretable machine learning in customer data analytics in wealth management. The proposed approaches and models are (1) MMDB for personality mining, (2) transfer learning for customer personality prediction, (3) ensemble model with text mining for churn prediction, (4) interpretable machine learning with SHAP-MRMR+ to extract customer insight, and (5) customer segmentation and managerial implications with personality and SOM.
From the financial aspect, this is one of the first research to utilize deep learning for socially responsible investment. The proposed framework consists of (1) text mining of CSR reports for ESG ratings, (2) ESG-based quantitative models, (3) deep learning using Multivariate BiLSTM for stock return prediction, (4) MV-ESG for ESG-based portfolio optimization, and (5) reinforcement learning for socially responsible investment.
The empirical results show the advantages and effectiveness of deep learning algorithms and big data analytics in financial wealth data analytics. Through the completion of this thesis, various aspects of wealth data analytics have been researched and integrated into sophisticated frameworks, and the information systems can provide meaningful insights for multiple stakeholders, from researchers to individual investors and fund managers.
Please use this identifier to cite or link to this item: