Philip Guo (Phil Guo, Philip J. Guo, Philip Jia Guo, pgbovine)

MIT: A Life in Emails

7071

That's the total number of emails I sent from my MIT email account during my 5 years there as a student from 2001-2006. During the summer after my final semester, I was archiving my emails and suddenly got the urge to write some scripts to extract information from them such as time, date, and recipients. My goal was to find patterns in my emails that reflected my lifestyle at MIT. I wanted to only focus on the emails that I sent, rather than the emails I received, because those more accurately reflect my purposeful behavior (I have to be actively working on my computer at the time that I send emails, but I can receive emails at anytime). The fruits of my labors are these Excel graphs that tell a story of my life at MIT through my emails.

The Big Picture

Number of emails sent per half-month from 2001-2006 during my 5 academic years and all vacations (7071 emails total).

The number of emails that I sent gradually increased every year, with highs during the busiest times in each semester and lows during vacations such as summer, spring, and winter breaks. Check out the huge spike in emails during my final semester, mainly due to my TA duties for 6.170.

Number of emails sent per half-month from 2001-2006 during my 5 academic years and all vacations, overlapped to display the span of one academic year (7071 emails total).

This graph is like the previous one, except that I've taken the number of emails sent in all 5 academic years and stacked them on top of each other, with one color representing a particular year. This format better illustrates my emailing patterns throughout the academic year. The two lowest bars are in the second half of December and the first half of January, which is winter break. The summer months also have fairly low activity, with early July being the lowest and then increasing again in late August as I prepare to start a new school year. In general, there is higher activity during the spring semester than the fall, which is consistent with me doing more in the spring in terms of both work and extracurricular activities. The fall semester starts off with lots of action at the beginning of the term and then tapers off slowly. The end of October and beginning of December—mid-term and pre-finals, respectively—show spikes in activity. The spring semester has two humps of heavy action separated by a local minimum during spring break (the second half of March), the first at the start of the semester in early February and the second right after I return to school from spring break in early April.

Work and Play

In the following graphs, I only count the emails that I sent while on campus at MIT, so I have filtered out emails sent when I was at home, on vacation, etc. I have grouped them together by semester (10 total).

Number of emails sent per semester for 5 academic years when I was on-campus at MIT during 2001-2006 (5938 emails total).

In general, the number of emails I send monotonically increases over time, with the overwhelmingly large peak being in my final semester (M.Eng. Spring). The green bar shows the enormous number of emails I sent as part of my TA duties, but even if that were omitted, the M.Eng. Spring semester would still have the most sent emails. The blue bars show personal emails I sent to my friends and close acquaintances, the red bars show emails that I sent to my research colleagues, and the yellow bars are for all other emails. In my freshman year, I sent very few personal emails because I didn't have too many friends at that time, and most of my emails were related to classes and other administrivia. I got involved in undergraduate research during the beginning of my sophomore year, and my involvement grew over time, as evidenced by the increasing size of the red bars.

Percentage of emails classified as Personal, Research, and Other (excluding 6.170 TAing during M.Eng. Spring semester)

This graph takes the previous one, cuts out the green TA bar, and stretches all other bars vertically in order to display the percentage of emails sent to a particular category of people during each semester. It's more evident now that during freshman year, most of my emails were 'Other', but the amount of 'Other' emails gradually drops over time. By my senior year, 2/3 of all my emails are either for personal or research, which makes sense because most of the time, I would either be working on research or emailing my friends. I took very few classes during my senior and M.Eng. years in order to dedicate more time to research, and I wasn't applying for jobs or involved in other extracurricular activities, so the amount of 'Other' emails dwindled considerably. The number of personal emails peaked during my junior and senior years, because those were my most social years. It dropped during M.Eng. Fall because I was so busy working on grad. school applications, and rose again in M.Eng. Spring because I was more social during my final semester.

The Flow of the Weekdays

Percent of emails sent per weekday for 5 academic years when I was on-campus at MIT during 2001-2006 (5938 emails total).

Unsurprisingly, most of my emails were sent during the weekdays, but a decent amount were also sent on Sundays, because that was my main work day during the weekends. As a student, I worked pretty much 7 days a week, with the exception of many Friday and Saturday nights (I spent the least amount of time in front of the computer on Fridays and Saturdays, because those were my most social days of the week). In contrast, I would expect the email patterns of someone working an office job to be mostly concentrated on the weekdays with very few emails on the weekends.

The two most interesting sets of bars are the ones for Junior year (blue) and the ones for M.Eng. year (yellow). Junior year was the most indicative of my undergraduate years, because I spent most of my time on classes and did research only about 10-12 hours a week. The heights of the blue bars peak on Monday and then monotonically decrease throughout the weekdays, and then rise again on Sunday. This pattern is indicative of my undergrad life where there is heavy email communication early in the week to get myself oriented on new assignments, and then things level off throughout the week as I focus on working on assignments. (The same pattern is also reflected in the bars for Sophomore and Senior years.) In contrast, the heights of the gold bars of M.Eng. year show a rise and hump on Wednesdays (hump day!), then a fall in activity. This pattern is indicative of a typical work week in the real world, which makes sense because, during my M.Eng. year, I was pretty much working on research full time in the office (30-40 hours a week). In sum, the contrast between the blue and gold bars shows the difference between email patterns in undergraduate life (heavy on classes) and graduate life (heavy on research).

What times of the day was I working on research?

Short answer: almost all the time! Long answer ... see graphs below:

Number of emails sent to primary research colleagues at the Program Analysis Group from 2004-2006 (818 emails total). Morning is 6am-1pm, afternoon is 1pm-7pm, and evening is 7pm-6am.

As evidenced by my email sending behavior, I did most of my research work during the afternoons. I was not much of a morning person; I always woke up sometime between 10:00 or 11:00 and then spent my mornings running errands or being unproductive until lunchtime. Working on research is definitely not a 9:00-5:00 job, though; I worked a lot during the evenings as well. Notice the lowest bar for evening emails in the above graph was on Friday evenings, because I usually make it a point to go out with friends instead of working in the office (I also tried to avoid working on Saturday evenings if I could). Sunday mornings were also quite lazy, because I didn't usually get back in the groove of working until the afternoon and evening.

Number of emails sent to primary research colleagues at the Program Analysis Group from 2004-2006 (818 emails total). Notice the color coding for morning (6am-1pm), afternoon (1pm-7pm), and evening (7pm-6am).

The above graph shows a 24-hour view of the emails that I sent to my research colleagues. Notice that my mornings usually didn't start until 10am, and most of my emails were sent from 11:00am-7:00pm, when I was physically present in the office. However, work doesn't end after I leave the office! I would usually continue working long into the evenings after I returned back to my apartment, usually until 1:00am. I usually slept from 2:00am to 10:00am, as evidenced by the lull of email activity during those times.

Number of emails sent per hour when I was on-campus at MIT during 2001-2006 (5938 emails total). Notice the color coding for morning (6am-1pm), afternoon (1pm-7pm), and evening (7pm-6am).

This final graph shows when I sent all emails (not just research emails) when I was on campus at MIT during all my 5 years. There is a lull in email activity during lunchtime (~12pm or 1pm) and dinnertime (~7pm or 8pm), but otherwise, there is heavy email activity during the afternoons and evening, even late into the night like 1am or 2am.

Conclusion

Unfortunately, the results I obtained from my fairly crude and superficial analyses (mostly making histograms according to dates and times) weren't as dramatic or spectacular as I had originally hoped for, but that's a risk of exploratory research ... you're not guaranteed to get fascinating results. I ran out of motivation before exploring more sophisticated statistical analyses or even data clustering, but oh well ... this was just an impulsive little side project, and I learned some skills in the process of doing it, so it's all good.

Created: 2006-08-28
Last modified: 2006-09-24