I really like the idea behind the Google Glass project but from the beginning I was concerned about the future of privacy, if these devices are around us. Mark Hurst wrote an excellent article about this topic.

Now, I know the response: “I’m recorded by security cameras all day, it doesn’t bother me, what’s the difference?” Hear me out – I’m not done. What makes Glass so unique is that it’s a Google project. And Google has the capacity to combine Glass with other technologies it owns.

and

The most important Google Glass experience is not the user experience – it’s the experience of everyone else. The experience of being a citizen, in public, is about to change.

That’s the point.

A deeper look into all of my tweets (pt. III)

A few days ago I wrote about my bugfixing and now I had the time to go on a bit. I converted all my tweets into my new csv scheme. And of course again, I plotted a histrogram to see the distribution of tweets by their length.

With the old tweets it looked like this:

After my setback from last night, I updated my script, which converts the JSON files into CSV files. As far as I can see right now it works now the way I want it to work.

As I wrote here the output of the script before looked like

1
2
3
4
5
6
7
8

   id           timestamp                                     tweet
1 640010912 2008-01-25 12:58:37        dynamite deluxe genießen!!1 L:home
2 636932532 2008-01-24 18:06:34                      audimaxparty!! L:uni
3 635830622 2008-01-24 12:23:35                  chillen in der uni L:uni
4 631430752 2008-01-23 09:55:03 mal schaun wie sich pownce.com entwickelt
5 592152102 2008-01-12 22:01:24         gleich neue folge numb3rs schauen
6 590962562 2008-01-12 13:46:33              hausaufgaben machen L:zuhaus
...

And as I found out yesterday this sometimes even looks like

1

n 11111111 2000-01-01 12:12:12 RT @longusername: this tweet has now more than 140 chars, because of the retweet part added at the beginn...

Tweets were cut off at the end, because “@RT username:” was added at the beginning. Fortunately the original message was also written in the JSON files, so I could extract it. The new CSV files have also two more columns, isRetweet and author, which are self-explanatory.

So, now I can go on with my analysis. A new post will be published in the next days.

Setback

Last time, I showed how to load your downloaded tweets into R and sanitize them into a nice format.

This time I will focus on some basic statistical analysis of my tweets. We start with some basic information about the lengths of my tweets.

Just remember, the data frame with the name tweets looks like this:

1
2
3
4
5
6
7
8

   id           timestamp                                     tweet
1 640010912 2008-01-25 12:58:37        dynamite deluxe genießen!!1 L:home
2 636932532 2008-01-24 18:06:34                      audimaxparty!! L:uni
3 635830622 2008-01-24 12:23:35                  chillen in der uni L:uni
4 631430752 2008-01-23 09:55:03 mal schaun wie sich pownce.com entwickelt
5 592152102 2008-01-12 22:01:24         gleich neue folge numb3rs schauen
6 590962562 2008-01-12 13:46:33              hausaufgaben machen L:zuhaus
...

1

summary(nchar(tweets$tweet))

If you do the same with your tweets you will get something that looks like the following:

So, what does this mean? Min. and Max. are self-explanatory - I’ve at least one tweet with length 1 and at least one with length 140. Interesting is the 3rd-quantile. I can derive from this, that 75% of my tweets are shorter than 97 characters.

In the last episode of Chaosradio (page is down at the moment but will be back soon) host @Holgi played some nice background music. He later tweeted the URL of the website where you can listen to this music. There are several mixtapes and so far I’ve only listened to the “Klänge der Nacht” series and I like it because it doesn’t distract me when I’m coding but it also has some relaxing effect on me. Take a look and listen to some mixtapes, maybe you like them too.

A deeper look into all of my tweets (pt. I)

A few days ago I finally got access to my Twitter archive. And since I’m playing around with R at the moment I thought, why not combine these two things. I read some articles about data analysis and data mining with R in the last weeks so I thought it would be a nice training to apply this to some data and maybe get some nice results.

Since there is so much you can do with thsi data, this will be the first article in a series of articles covering my approach.

hello, world!