My 2018 in data

It is the time of the year when you look back at the promises that you have been trying to keep but failed.

An honest review of the year 2018 using data collected via Fitbit, web browsing time via Chrome, and self-logged time.

Breakdown of awake time in an average day

time logged

This time usage data is manually logged using an IOS app, for about three months during summer and fall. Categories do have some overlap, and measurement error is inevitable. But it is mostly an honest, relatively accurate and representative sample of my average time use during both weekday and weekends.

Of all the time awake, only 40% time is dedicated to work and study. In term of hours, for an average of 7 hours sleep tracked by Fitbit, that is about 7 hours of actual work time every day. This excludes internet surfing time, time spent on eating snacks, chatting with friends and meeting with collaborators. In the future, I want to use the Pomodoro Technique to validate the measurement of focused time by counting the number of tomatoes achieved every day.

About 20% of the awake time is spent on “fun stuff,” amounts to 3.5 hours every day (probably more on the weekends). 12% of the time on spent on browsing information online and reading news or articles, which should have been working time.

Around 12% of the awake time is spent on preparing and eating meals, about 2 hours every day. That shows how serious I’m about food, LOL. On the contrary, less than 1% is spent on working out. This certainly explains a lot about my lifestyle in these months.

Breakdown of web time usage

web time

The web time tracker is automatically tracked by a Chrome extension since 2018-09-18 ( 106 days in total including 85 active days ).

Only 27% of web browsing time is work-related, 28% of the time is spent on video sites like Youtube (tech and cooking channels) and less than 10% on reading.

In absolute terms, I spend about 8 minutes every day on Youtube and other video websites, 8 minutes on work, 1 minute and a half on reading news, and about 2 minutes on searching for random stuff.

Fitbit data

I wear the Fitbit charge HR since June 2017 and I just recently found that they do provide the raw data for each user to download on the fitbit website. Thanks to Fitbit and the Fitbit community for constantly demanding this functionality, this data creates an opportunity for every Fitbit user to experiment on their way of eating and living and see how that affect their health indicators. The data comes in JSON formats and has numerous features. I tried with a few of them, and some visualization of them are listed below. For obvious reasons, I removed my raw data, but you might find the code to process and plot useful in the Github repo of this website.

Rest heart rate

One of the easiest, and maybe most effective, measure of health is the rest heart rate (RHR). An RHR between 60 and 100 beats per minute is considered normal for adults. According to Harvard Health Blog, a 2013 study in the journal Heart finds that an RHR between 81 and 90 doubled the chance of death, while an RHR higher than 90 tripled it.

rest heart rate

The figure above is my RHR over the past year and a half. The black dots are the exact readings of the rest heart rate, while the vertical bars are the possible measurement errors (which are larger than I thought). It seems that all the readings are within the standard region with small fluctuations from months to months. Except for some extreme cases of measurement error, the RHR is still far from the 80 danger zone. But it also serves as a warning sign for me to work on my cardio to maintain a lower rest heart rate.


Fitbit has an excellent analysis of tracking movements during sleep and measures the duration and quality of sleep at night. My life is pretty regular and the sleeping hours remain relatively stable over time. I was interested in whether my sleep patterns change across months/seasons, or behave differently on weekends.

sleep month

I chose two variables: average minutes asleep per night and the average sleeping efficiency ( which is a measure of overall sleeping quality). The seasonal trends are not obvious, except for that I struggle more to sleep during summer (more restless during sleep or takes longer to fall asleep ). I sleep a lot less during December, but that is because I usually travel back to China during winter break and there were a lot more complications involved.

sleep weekends

For the weekday/weekends comparison, the mean of the data is very close between the two categories. But weekends data seem to have more variations in both length and quality of sleep. Some possible explanations might be the difference in the amount of alcohol consumed, pressure from work and physical activities. But I will need to conduct a more controlled experiment (or more disciplined life ) to make a more convincing argument on weekdays vs. weekends sleeping patterns.


step month

My steps are significantly lower during the winter months when it is colder, but not a lot more in the summer except for a few time when I go hiking or traveling.

step daily

The day to day variation on this daily plot is too much to tell something meaningful.

To put numbers into more meaningful terms, I used two benchmark numbers.

10,000 steps, which is approximately five miles, is a number said to help reduce certain health conditions, such as high blood pressure and heart disease.

80.98% of my days are below this number.

Less than 5,000 steps per day are considered inactive, which is 34.42% of my days.

That being said, steps are not perfect measures. HIIT, and lifting weights probably won’t accumulate many steps per se, but they have proven to increase heart rate rapidly and burn a lot of calories within relatively a short duration of time. Fitbit does automatically record workout and active minutes. I will try to record more and analyze them in the future, including their association with changes in rest heart rate, changes in weights and body fat.

Achieved results:

1 paper submitted

1 certificate earned (Data Science mastery)

2 websites built (personal webpage and a BI platform)

3 news articles published on trade

5 data science projects (and a significant amount of work on organizing the code and Github repository)

6 blog posts (way to go on the writing plan)

Goals for 2019:

Less time spent on videos and entertainment, more steps and workout time!

Track more data on health and time use, manage for life through data.

Yujun Zhou
Data Scientist

Currently works at Facebook Applied AI Research as a Data Scientist. Trained in Applied Economics and Machine Learning.