This repository contains all files with which the analysis the Bachelor thesis regarding microtransactions in Bitcoin was conducted.
Publishing every line of code is an attempt at maximum transparency (which methods were used? Did conceptual or implementation errors lead to wrong conclusions?) and reproducability (running the code in the repository should lead to the exact same data-set and Statistical results my research is based on).
Using git for version control also serves as a universal "undo button" throughout the entire project.
The Javascript files were tested with node v5.2.0 available here:
The code relies on features from ES2015, such as lexical scope via let, arrow functions, Promises and template strings, thus a modern version of node.js is needed.
install via npm install lodash sqlite3
The data-source for Bitcoin transactions, namely timestamps, transaction-hashes, volumes transacted and fees are from pulled Coinbase's Toshi API remotely:
Data about network utilization (Daily averages of block-size) is provided via verison 0.11.2 of bitcoin-core.
In order to use the full API of bitcoin-core, bitcoind (or bitcoin-qt, which in turn invokes bitcoind) have to be run with the flags "-reindex" and "-txindex" after the local blockchain is in sync with the overall network.
The JSON-RPC-API is queried via the node-bitcoin package, available here:
NPM bitcoin package,
install it via npm install bitcoin.
The results from the JS-scripts are written into a local SQLite instance via the sqlite3 package (https://www.npmjs.com/package/sqlite3). The idea here is to be able to run SQL queries on the large dataset.
The aforementioned queries are executed out of R within R-Studio, to get the results in R's native format. The R-package supplying this functionality is sqldf, availabe here: https://github.com/ggrothendieck/sqldf