Twitter recently allowed users to download their twitter archive by following these directions. It is a pleasant surprise that within the archive files is a sweet CSV file with every tweet from the account whose archive is requested.

The @KenyaOpenData has over 3000 tweets and should make some interesting analysis. After downloading the entire archive, a simple histogram that shows the frequency of tweets can be developed as below.

Rplot01

It is immediatelly apparrent that the frequency of tweeting is intermittent with the obvious outlier being May 2013. Further investigations reveal that Thursday, 30th May 2015 had over 600 tweets on #OpenGovAfrica which is likely to affect all time and day related analysis.

timestamp text
2013-05-30 14:20:38 Thanking our partners @kenyaICTboard, @ICJKenya, @Ushahidi, @Open_institute, @TIKenya, @AfriCOG #OpenGovAfrica
see http://t.co/zel068 pZyS
2013-05-30 14:18:06 Special thanks: Daudi Were @mentalacrobatic Robert Hunja @RRHunja Paul Kukubo @pkukubo Serena Hotels @SerenaHotels Bitange Ndemo @bantigito
2013-05-30 14:13:34 Thanks to the international participants from Nigeria, Liberia, South Africa, Ethiopia, UK, Ghana, Belgium, Tanzania, Malawi #OpenGovAfrica
2013-05-30 14:11:28 See more photos from the #OpenGovAfrica outreach meeting facebook page – https://t.co/S7lIYDOX5w
2013-05-30 14:10:18 Thanking all #KOT for making #OpenGovAfrica a fruitful on-line discussion. Thank you all for your contributions!
2013-05-30 14:08:28 We thank all that made this year’s #OpenGovAfrica outreach meeting a great success. Many thanks to our sponsors and partners!

It can also be seen that the first tweets were sent on 8th Dec 2011 at 17:25. The next four tweets can be seen below.

timestamp text
2011-12-08 17:25:26 This is the official Kenya Open Data Twitter account. For information on the Kenya Open Data Initiative!
2011-12-08 17:29:29 RT @g33kmate: I hear that the Kenya Open Data initiative is on Twitter and its official handle is @KenyaOpenData follow and RT for more …
2011-12-08 17:30:36 @Juuchini thank you. Let us get more follow traffic for a more informed data process.
2011-12-08 20:00:30 «@kaburo Check out latest #ushahidi initiative targeting #harassment on streets of #Ramallah http://t.co/NSGrygeu #Palestine @CafeThawra»
2011-12-09 10:21:20 Did you know that you can now suggest a dataset On the Open Data Portal? http://t.co/49uSsDYJ. What data would you like to see?

It should therefore be interesting to find the number of tweets we have sent out on annual basis.

Rplot

The year 2012 has the most tweets which shows a general fascination post creation of the account in December of 2011. Additionally, the frequency of tweeting has generally gone down since 2012 has the majority of the tweets despite a cumulative number of 600 tweets contributed on the 30th of May 2013. The year 2014 has the least tweets.

What pattern does the daily tweeting habits of @KenyaOpenData form over a week?

Rplot02

Admittedly, this is a very ‘normal’ histogram which follows the ‘bell curve’ as expected. Despite the expectations that we would have more tweets on thursday influenced by the 30th May 2013’s #OpenGovAfrica, Wednesday leads in the number of tweets in a week. Research identifies that there are usually very few tweets over a weekend and these data proves as much. Tweets on Saturday and Sunday are a testament to the dedication of the @KenyaOpenData’s Staff. Ok that was little too much!

Twitter Amageddon: Which months do we go all out?

Rplot03

Evidently, #OpenGovAfrica has a huge influence on the number of tweets in May and therefore makes it an outlier. Otherwise June and October are the two peak months. The months of September to November have almost equal number of tweets.

Tweets by Time of Day: Twitter all Day long?

Rplot04

There are mid morning and mid afternoon peaks in the twitter activity and slumps everytime the boss shows up. Hehe. Isn’t it always easy to switch to a different tab on the browser just because the boss walks in even if twitter is part of the delivarables? Ok looks like that doesn’t happen at @KenyaOpenData or the boss never walks in. It could also mean the boss herself is twitting!

It is very concerning that someone tweets in the middle of the night! Who’s he(of course it’s guy. The ladies in the team don’t have insomnia) trying to reach? Big Foot?

What might be going on at night?

Rplot05

It is difficult to understand the night owl at all. There is no apparrent trend in how he operates. Are they drunk tweets? Ok maybe that’s a logn bet.

How many of @KenyaOpenData tweets have hashtags?

Rplot06

More than half of the tweets are with hashtags which can be a show of the number of events we support and organise.

Does @KenyaOpenData retweet more or send out original tweets

Rplot07

Evidently, there are more original tweets that retweets. ### Does @KenyaOpenData reply to your tweets or just send out original tweets.

Rplot08

Admittedly, it was expected that there would be more original tweets that relies but not at the scale shown. It can be argued that there are very few questions that demand the attention of @KenyaOpenData which can be supported by the overzealous #OpenDataKE community. It can also be argued that there are very few conversations that @KenyaOpenData can join and contribute new insights which implies that the community is very knowledgeable.

What is the proportion of regular, RT and Reply tweets.

Rplot09

This is a proportional graph and therefore it helps find the distribution of tweets be they original tweets, retweets and replies. Evidently, there are very few replies compared to original tweets and retweets. This probably means that @KenyaOpenData is now more likely to reply to tweets than before, a phenomenon economists call the “over compensations effect”.

Twitter’s 140 character limit: Character for Sale?

Does @KenyaOpenData always use the limited 140 characters or are there times there are a lot more to spare?

Rplot10

So anyone who needs extra characters for their tweets, hit @KenyaOpenData up because they don’t have much to say anyway.

Open Data on Twitter

Prestone Adie


I drive Data Analytics where I surface stories from data that might not be immediately obvious. With a background in Actuarial Science I'm proficient in R and Python for data analysis and takes avid interest in anything data. Find me on twitter Follow @AdiePrestone where we can talk data, street food, cars and books.


Post navigation


Leave a Reply