Hadoop implementation of tweet length and most common hashtag analysis. This was a coursework assignment based on Twitter data collected during the 2016 Olympics in Rio.
Assumes that the tweet input file contains semicolon separated values with the following columns:
epoch_time;tweetId;tweet(including #hashtags);device
For example:
1530473325000;123456789;I like big #dogs;<a href="http://twitter.com/download/iphone" rel="nofollow">Twitter for iPhone</a>
The athletes input file is the one collected from https://www.kaggle.com/rio2016/olympic-games/data.
Generates a distribution of tweet lengths, grouped into buckets of 5 (for use in histogram plots). Full UTF-8 support for length calculation is included, just like Twitter's own implementation.
Computes the number of tweets that occurred during each hour in the dataset. It will generate at most 24 outputs (days are not respected).
For the most popular hour found in PartB1Job
, emits and saves to HDFS the top 10 hashtags tweeted
during that hour. Full UTF-8 support for hashtag parsing is included, just like Twitter's own implementation.
Computes the top 30 athletes from the Kaggle dataset, based on the number of mentions. Includes the number of mentions in the output.
Computes the top 20 sports from the Kaggle dataset, based on the number of mentions. Includes the number of mentions in the output.