BitcoinUses

Bitcoin Blockchain Analysis

Results

• All our results are live on our website
• The following is a stacked column chart made with Excel from the raw data:

• Raw results are in results.csv

Procedure

• The following diagram illustrates the whole process

Note that coinbase transactions, change outputs and transaction with a single output are ignored

1. Each mapper starts by obtaining historical bitcoin price information
   • Information is obtained from blockchain.info
   • A weekly average price is computed from the data, since 3 January 2009
2. Each bitcoin block is fed to one of the mappers
   • The mapper outputs the week number matching the timestamp of the block as the key
   • The mapper outputs a map of the counts of outputs for different ranges of money (USD) in the block as the value
     • Coinbase transactions are ignored
     • Transactions with a single output are ignored
     • Change output (largest one in transaction with multiple outputs) is ignored for every transaction
3. The key, value pairs from the mappers are fed to the reducer
   • The reducer simply sums all the values for each week
   • The final output is the key - value pair where
     • key: week number since 3 January 2009
     • value: counts of outputs per USD value range

Run it

1. HDFS setup: Refer to the Data ingestion section
2. Program build
   • Make sure you have at least Java JDK 1.7 installed; check with:

       java -version
     

   • Make sure you have Maven installed; check with:

       mvn -version
     

   • In this directory, enter this command to build the program

       mvn clean install
     

3. Program run
   • In your hadoop environment, enter:

       hadoop jar target/blockchain-1.jar /user/cloudera/bitcoin/input /user/cloudera/bitcoin/output
     

   • Check the results with:

       hdfs dfs -cat /user/cloudera/bitcoin/output/part-r-00000