I am using the twitter data from 2/01/17 in order to compare against previous lab results and to stay consistent. When sorting out the data by user_lang, I noticed a decent amount of languages used, being bg, da, de, en, en-gb, es, fr, id, it, ja, lv, nl, pl ro, ru, and su. After running the countif function, I have a total of 4450 tweets in English. Out of a total of 4790 tweets, the percentage of tweets in English is 93%.
I am not surprised most of my data is in English because this topic is related only to the United States, a primarily English speaking country. I noticed that most of the Russian tweets were retweets translated by the same twitter user named Slava381977. There were very few French tweets, only 7 to be exact, and the other tweets account for less than 1% of my twitter data, so it is safe to assume most of my tweets are in English.
Since my selection of data was only during a period of a day, I broke down the tweets into the number of tweets per hour. The most prominent times for people to be tweeting are around 5AM and 10PM. I am not sure why there were no recorded tweets from 7 AM to 1PM, it could have been an issue with the collection of tweets where I cut off the data stream prematurely. I understand why there was such heavy use at 10PM, because during this time the television show Sean Hannity is on where he discusses very controversial topics. The data is from when U.S. Attorney General Jeff Sessions was nominated for his current position, and it was a very controversial nomination since he is very pro-Second Amendment rights. Fox &Friends is on tv from 4-6AM, so this could be the possible reason for a heavy number of tweets at 5AM. Another reason for heavy traffic between 5-6AM is this could be when most people wake up and they want to tweet about something. Aside from the missing data, the flow of tweets seems common, with not a lot of tweets in the middle of the day, and heavy use in the morning and night.
The data above only represents hours when data was collected, meaning the period where no tweets were recorded are not included. Having an average of 282 tweets per hour is great because it is not far from the median of 286, meaning there are not too many outliers in my data. This is great because it means the data is consistent throughout the day, even during peak and downtimes. I was surprised to have a mode value since it is rare for there to be the same amount of tweets in two different hours. Having the same median and mode is interesting, with the most common value being the same as the exact middle value. My biggest takeaway from these numbers is my data is consistent rather than a large amount being collected in a small amount of time. Since my tweets are being sorted by hour instead of by day, it is tough to compare my tweets to the class data.
Having a max of 391 and min of 169 helps show the consistency of collection of my data since it shows the downtime of collection still had a reasonable amount of tweets, and the peak was not a considerable increase. with a range less than the average at 222, helping to show the consistency of my data. One thing I have noticed is the majority of tweets in Russian occur from 1AM to 5AM, showing there was no real downtime for my hashtag. It is surprising to see the quantity of tweets from Russia relating to my hashtag since it is regarded a United States issue rather than a global issue. The reason for the tweets could be for the same reason stated above regarding Jeff Sessions nomination.
Harrison, I found it interesting how you related your data to what time in the day did the (conservative) media discussed the gun issues. I did not know how important conservative figures such as Sean Hannity had on the debate. Do you know what Sean Hannity and Fox&Friends actually said on their broadcasts on those days?
I am not surprised by the percentage of English tweets, but I am surprised by the number of Russian tweets. I searched “2a Russia” and found an awards site – perhaps the award show happened or would happen soon during your data collection. Russians could have tweeted about gun rights – do you know what the gun control laws are in Russia? Are they similar to the US? What is Putin’s standpoint on guns? I also think that you should keep in mind that there is probably a #2a equivalent hashtag in different countries because #2a relates to a very specific section in the U.S. Constitution, not to the French or British or Canadian or any other constitution. If you have time, you should find the #2a equivalent hashtag of another country and see if there are similar trends to your data.
Your data made me think of the very detailed factors that could influence my data. I have to now see what liberal media do people who tweet #nobannowall watch and see if it influences or adjusts my data in any way.
A’arry, my boy… that was my best Hagrid impression. Anyways, congrats on completing another lab. Tiny steps will always take us towards our goal. I think it was very cool that you used the length of a day to take a magnified look at the mini rises and falls of the tweet trackers. Very thoughtful of you to consider what shows are on tv and how they might make people think (and tweet). Maybe we can look into what was said by Hannity that day so we can know what triggered the change. I really think that looking at one day like this makes me see data very differently! There is so much data that is out there, it is waiting to be explored!
have a great weekend, my lord!