Blockchain and Cryptocurrencies
Measuring indicators of economic activity in the Cryptocurrency and Token Transaction Networks.
Several cryptocurrencies (among which Bitcoin is the most apparent example) have been widely adopted in the past years. These systems show in interesting dichotomy: while they are intended to be decentralised, they are designed following rigid technocratic approaches. Interestingly, they constitute closed economies where individual transactions can be traced with large levels detail: While anonymity is preserved by different means, all economic activity by the users is recorded in public ledgers. In this proposed work, the student will study indicators of endogenous economic activity (signalled by both: the economic transactions between users, and capital accumulation) and compare it across digital currencies or tokens. This family of topics are suited for students with a strong interest in data science and/or economics.
Start of the thesis: anytime; required knowledge in R or Python: BA: basic, MA: upper-intermediate; type of work: BA: theoretical, MA: theoretical and empirical.
Technical Analysis of Cryptocurrencies Using Functional Data Analysis
How does a technical glitch or the introduction of a new policy impact the volume of transactions as well as the price of cryptocurrencies? Is there an acceleration of the trading after such an exogenous shock? Can we observe different clusters of cryptocurrencies in terms of trade volume or price?
If these questions can be answered with usual time series techniques, this thesis should
apply Functional Data Analysis (FDA) and Functional Principal Components Analysis (FPCA) to answer them. Functional data analysis is increasingly being used to better analyse, model and predict time series data. Key aspects of FDA include the choice of smoothing technique, data reduction, adjustment for clustering, functional linear modelling and forecasting methods.
Start of the thesis: anytime; Required knowledge in R or Python: MA: upper-intermediate; type of work: empirical.
Reconstructing hidden social connections between customers in a large physical market
Dealing with incomplete data challenges our ability to leverage social networks of customers for marketing purposes. In the best-case scenario, marketing researchers may have access to data on customers’ communication patterns relative to a specific communication channel (e.g. phone calls and/or sms) and provided by a specific company (e.g., mobile phone operator). Even in this ideal scenario, one would lack information on communication flows that took place through alternative communication channels (e.g., Whatsapp), and between individuals who are not customers of the company. In this thesis, the candidate will analyze a large communication network from a mobile phone operator in a developing country, provided by an ongoing collaboration with a US-based marketing company. In this unique dataset, the candidate will have access to the communication flows between the company’s customers, to the communication flows between customers and individuals who are not customers ("non-customers"), and to the time-series of credit-card purchases for some of the customers and non-customers. The main goal is to develop and validate algorithms to reconstruct the social connections between the non-customers. Two classes of algorithms will be considered: (1) structure-based algorithms that analyze the available communication flows to infer missing connections; (2) dynamics-based algorithms that take into account the similarity between the individuals’ purchasing patterns in shops. To achieve an improved accuracy, both types of algorithms will be complemented with individuals’ metadata (e.g., age, income, etc.). The first step of the thesis will consist in testing existing algorithms; the understanding of the relative performance of existing methods will likely lead the candidate to the design of improved algorithms.
Master thesis. Start of the thesis: anytime; required knowledge in R (or equivalent programming language): intermediate; type of work: theoretical.