M.A. Theses
Permanent URI for this collection
Browse
Browsing M.A. Theses by Subject "Bitcoin."
Now showing 1 - 2 of 2
Results Per Page
Sort Options
Item Consumer adoption of Bitcoin in e-commerce(Thesis (M.A.) - Bogazici University. Institute for Graduate Studies in the Social Sciences, 2019., 2019.) Serhatlı, Serhat.; Onay, Ceylan.Bitcoin has recently brought disruptive innovation into the financial system. While it is considered as an investment instrument by many, its adoption as a payment mechanism particularly in e-commerce has been scarce. Nevertheless, many e-commerce sites are considering or experimenting with it as a payment mechanism. However, the content analysis made in this thesis to identify trends and research gaps in this field shows that academic literature on the perceived benefits and risks of cryptocurrency adoption is limited. In fact, there is no prior study about cryptocurrency adoption in e-commerce payments. Accordingly, this thesis aims to explore cryptocurrency adoption in the e-commerce context to fill this gap. The cryptocurrency adoption in the e-commerce context is explored through the diffusion of innovation theory, which is expanded with perceived security, ubiquity, financial risk, and the legal risk constructs. Data is collected from 208 respondents via an online survey with convenience sampling. Frequency, crosstab, reliability, confirmatory factor analyses and partial least square estimations were conducted over the data. The results show that relative advantage, compatibility and perceived security significantly explain cryptocurrency adoption in e-commerce. In addition to that, ubiquity and trialability were significant predictors of perceived security. Finally, managerial implications are discussed for designing customer experience in cryptocurrency payments.Item Predicting the Bitcoin trend using technical indicators for deep learning algorithmic features(Thesis (M.A.) - Bogazici University. Institute for Graduate Studies in the Social Sciences, 2019., 2019.) İnce, Nuri Tuğkan.; Durahim, Ahmet Onur.This research focused on utilization of technical indicators for predicting the trend of Bitcoin/USD price with different deep learning algorithms. The goal was to achieve better trend prediction accuracy results using technical indicators compared to using only close, open, high, low and volume (OHLCV) data for Bitcoin/USD parity. Through achieving this goal, three different deep learning algorithms, Deep Neural Networks (DNN), Long Short-Term Memory (LSTM) and Gated Recurrent Units (GRU) were used because of the performance they exhibit in literature for financial stock prediction domain and their theoretical convenience. 156 technical indicators, mathematical transformations and financial patterns were used in feature set to test against OHLCV data of Bitcoin/USD. Experiments in this research showed that, utilization of technical indicators produced better accuracy results compared to OHLCV data for all three prediction models. For the imbalanced dataset distribution produced by a one-way transaction cost to decide buy, hold or sell operations, LSTM performed best among the models used in this research with achieving 56.33% accuracy score with reasonable individual class prediction rates whereas raw data could achieve 53.26%. In the scenario for which the one-way transaction cost is tuned to have a uniformly distributed dataset, GRU performed best with achieving 52.19% accuracy score whereas raw data could achieve 39.85%.