Twitter can be used to gauge the impact of certain topics and the effectiveness of certain add campaigns. Many advertising agencies and political consulting firms relay on the mining of twitter to calculate the sentiment of marketing campaigns. Due to twitters robust api it is possible for individuals to effectively search for trending topics and analyze them for effectiveness. An important topic currently in American politics considers the affordable care act. This is new healthcare law in the united states that changes the requirements and laws regarding the purchase and administration of health insurance. Many different opinions exist about the implementation of this law and a conversation has been brewing on twitter because of it. By taking a snapshot of this conversion we can gain an initial understanding about this conversation.
The twitter api was accessed using the search function to gather 1000 tweets related to a particular query. The three query used were healthcare.gov, obamacare, and affordablecareact, these terms were selected due to them being most related to the healthcare law. Each one of the tweets was filtered by screen name, hashtag, and text and analysis was performed on these corpora. Screen names and hashtags were extracted from the data and analyzed using basic summary statistics. Text analysis was done using python's nltk library to remove many of the non-relavant terms so the analysis will be more indicative of meaning of the texts. All source code is available on github.
Visualization of the text data indicates a power law distribution, more specifically a Zipf distribution, which is to be expected from text data.
The most used term in the text data is the "Affordable Care Act" and overall the text data most probably indicates a more neutral bias from the terms with the top twenty percentage usage. This is shown in the LogLog plot below.
Screen names follow a power law distribution that is typical to data twitter analysis. Affordable care act is by far the most used screen name followed by a distance second by obamacare. Also notable is the screen name tcot, which stands for Top Conservative on Twitter. This is most probably an indication of negative reaction to the law, however more analysis must be taken to come to a more confident conclusion.
Hashtag information may indicate how popular a particular item is because these hastags are created to be shared with other users. Obamacare leads the hash tags implementation, also the tcot hashtags and a @weknowwhatsbest hashtag exists which is a conservative pundit on twitter. The top twenty hashtags are below.
##Conclusions and Further Actions
These simple analysis indicate that as expected the affordable care act elicits a wide spectrum of
reactions from the American public. It also demonstrates how powerful a tool twitter is for sentiment
analysis. Further analysis must be done to get a better understanding of the sentiment. For example,
tweets could be collected over a longer period of time be setting up a remote server and taping the 1%
firehose.
While twitter is a great tool for topical analysis its limitations must be understood give any analysis
proper context. Selection bias is very possible not only due to the relatively limited usage of twitter
but also the power law distribution of active tweeters.