bitcoin-analysis

ECE 4813 cloud computing project for bitcoin price analysis and prediction.

Quick Start Guide

1. Clone this repository to a machine with Storm server already installed and running.

2. cd storm-server to switch to the storm-server directory.

3. make init to install the necessary python libraries and create a config.json.

4. Create the following tables in DynamoDB:

   Table Name	Partition Key	Sort Key
   bitcoin_analysis	date (String)	timestamp (String)
   bitcoin_price	timestamp (String)
   bitcoin_price_prediction	timestamp (String)
   tweet_sentiment	date (String)	timestamp (String)
   news_sentiment	date (String)	timestamp (String)
   reddit_sentiment	date (String)	title (String)

5. Fill in the config.json with appropriate data. This contains configuration information for the following:

   • Twitter API Keys
   • AWS DynamoDB Access Keys
   • Kafka Queue Names
   • Reddit API Keys
   • Google News API Key

6. make build to build the Storm *.jar file to run.

7. make deploy to deploy the build to Storm (this will automatically kill an existing deployment before deploying a new one)

8. make listen to start the listeners that will feed data into the Kafka queues.

View logs

Logs are extremely useful when Storm does not work as expected. They can be found in the following directory:

/var/log/storm/workers-artifacts/{topology_name}/{worker_id}/worker.log

topology_name and worker_id depends on the particular instance of the Storm server.

Run web server

in /frondend/ folder :

9. use make web to copy config file to current dir and start the web server. the default host is the public ip of ec2 instance

host and port can be tweaked.

default is "0,0,0,0" with port 80