Social media analytics and what it can tell us

Economics, Politics Pixelics's picturePixelics

Regular readers will know that until recently we included a "social chatter" series alongside some of our standard indicators. We have since removed "social chatter" entirely in favour of unique stand-alone social media indicators, which focus on deviations in the long-run trends of social media activity on a particular subject. We are calling these indicators "Social insights", and they will appear on our indicator page as they are released. All of our social insight indicators will be populated with fresh data at the end of every week, and each trend is derived from a minimum of 150,000 rows of data spanning multiple years.

Our first such indicator, released earlier today, is called Social insights: Australian housing.

social-insights-aus-housing.png#asset:6575

This indicator measures total Twitter user activity related to the Australian property market since 2013. The idea is that the more people talk about a subject, the more likely that something has either just changed or is about to change. We take great care to ensure that our data are as accurate as possible, e.g. by controlling for the number of active Twitter users at a particular point in time, so that our trend is actually representative of interest in the topic not just a result of the growth in total Twitter users. This is just the beginning and in the near future we plan to not only provide additional social insight indicators on a diverse array of subjects, but to incorporate tools such as sentiment analysis in them as well.

Please note, however, that given the experimental nature of our social insight indicators they are designed, at most, to complement traditional data sources such as those from central bank or statistical agencies. They should not be used as evidence of a turning point in a particular sector or economy, nor for the basis of investment decisions.

About the author

Notes published by this user are not attributable to a particular author. This user's notes usually, but not always, address a specific Pixelics feature or technical issue as opposed to original research.