Plot users by first mention of keyword

[aronwc]

Using the ~120 users for which you've collected the timeline, compute the first day on which they used one of those keywords. Plot a histogram of these dates for all users. (e.g., x axis is date, y-axis is how many users first used one of those keywords on that day)

Point is to see if people became "Active" on different days.

[ElaineResende]

I collected the timeline for users who mentioned vemprarua(cometothestreet), foraPT(PT out - PT is the party which Dilma is in), foraDilma(Dilma out) in the streaming data set. The distribution of of first day of use of these keywords is in the plots below.

[aronwc]

Why do you think no one becomes active in days 2-12?

[ElaineResende]

Aron, I am sorry, the plots were not right. I have just fixed them. I got confused with the dates, but now they are fixed.

[aronwc]

OK, looks better. Was there another event around 9/1 that would cause the spike at the end?

[ElaineResende]

I don't believe so, I am going to check. I know about protests yesterday (9/7 our independence day).

[ElaineResende]

I am so sorry again, I realized that the plots are still not right because I was reading the files from the beginning to end, and they are in decreasing order of data (Aug/2015 -Jan/2015 e.g.), so I did the reading in reverse order now. I am changing the plots now, to be more understandable. An example of how the plot is right now:

[aronwc]

OK; this makes more sense. There is also a clear set of "early adopters" followed by a large set of "late adopters".

[ElaineResende]

I hope I got these plots right this time.

I plot by trimester this time, here we have number of users vs. days for dilma out:

January to March - March 15th was the day of one protest

April to June

July to September

These are for PT out:

January to March

April to June

July to September

These are for come to the street:

January to March

April to June

July to Sep