Social Media is becoming ever pervasive in modern culture. One of the simplest use cases of social data is as a "temperature check" for how a brand is performing. This talk offers a simple walk-through of how python can be used to take Social Data from its raw form and transform it into a usable visualization to help understand the market a company exists in.
This project focus on how social media data extracted about the fast food industry can simply and repeatably be turned into a system for analyzing market position. This market position can by proxy be used to derive strategy.
Data will be extracted from Twitter for a variety of brands that operate in the same market space. That data will be split into its individual words and using nltk it will be cast to its root word. A TFIDF (sci-kit or raw code) analysis will be used to identify which words are uniquely used to describe the industry. The data will then be topic modeled in gensim to further reduce the number of dimensions and to reduce the number of feature vectors that need to be tracked. Finally the data will be reduced into a 4-plot using Correspondence Analysis or MDS. While these are all independently simple the analysis itself is extensible enough to solve this problem in a variety of circumstances.
Slides available here: http://www.slideshare.net/PyData/social-media-brand-positioning-workflow-david-gerson