Much like my gym regime I spend the majority of my time reading about the appropriate exercises and diets rather than implementing them. It is very similar to my attempts at learning R, Python and Tableau. To my astonishment you actually have to experiment and put in the effort to get your rewards. This is therefore my first attempt at getting data from Twitter, manipulating it and presenting it in Tableau. I would appreciate any feedback on this. It has provided to be a great learning experience.
The topic that I wanted to look at was #Storm Frank which hit many towns in Ireland and England very hard. Living in an area which flooded in 2009 I was intrigued to see how times had changed and that Twitter was used as an effective source to share information and provide updates and images on what was happening.
The insights that I would take from the data are that they were correlated with the storm hitting the country which made sense. It was also interesting to see that the DisasterChannel had the greatest retweet impact, followed by InfidelPrince (interesting screen name with a link to the UKIP website). The number of mapping points was less than I had anticipated but even the few points tallied with locations where the storm was most prevalent (Ireland and England).
The key learning points that I would take are:
- The volume of mapping data points were much smaller than I anticipated. In the over 70,000 tweets that I managed to capture only a fraction had longitude and latitude positions. I plotted the datapoints in the visualisations for interest to see the dispersion but only 150 tweets had this information.
- Finally using reproducible code for R, and understanding to some degree the use of Notepad++ with R.
- Getting my Twitter authorisation right thanks to the site here
- Realising that the searchTwitter function only can get data for 7 days in the past, hence the fact that I am missing key data over a critical period.