Making the Most of my Time

2021/2022 in Data

In the college basketball world, we spend a ton of time reflecting externally, making sweeping predictions about Final Four teams in October, trying to seed teams perfectly before selection Sunday, and wanting our Big Board to have takes that age like wine. We must find a good balance in this evaluation, being able to look into the evaluator, ourselves. It is cliche to view this as the most important time to do so. The New Year not only provides us with a blank slate but an easy means of measuring data. 

January 1st, 2021

I was taking an extended Winter Break, having been home since mid-November, staying at my childhood home in Milwaukee. It was a weird time, in the middle of the pandemic, not being all too sure about when it was going to be over. Between classes at home, not seeing anyone outside of the grocery store workers and the depression that comes along with college, days blend. I was having difficulties separating the weeks apart, with a lot of time to myself, likely not helping my mental state. The turn of the calendar was the perfect means to start tracking my days. I wanted a better means of holding myself accountable for my time, and I figured if I had a better understanding of how I was spending that time, it would be easier to spend that time more efficiently. 

February 6th, 2021

I spend the evening scrolling Twitter after an evening of Summit League games. I slept for about six hours before placing my bets, showering, and locking into some college basketball in the heart of the season. I worked a bit on a better model I was revising, before taking a walk in the evening before another four hours of college basketball. This was the norm on most days when I didn’t have class.

March 20th, 2021
Issues in Simplicity

I didn’t spend 17 hours watching college basketball on this day, but with a look at my spreadsheet, it would seem so. When I started on my tracker on the first of the year, I didn’t know it was something I’d want to improve on continuously, which explained the simplicity in the design, categories, and duration of time recorded. 

December 31st, 2021

I would usually go to sleep around midnight, walk for two hours, and do something college basketball related from 11 onward. I knew this was a poor representation of my time, and the fact that time spent watching draft film and bracketology were treated the same way in my spreadsheet would be an injustice to the true use of my time. 

January 1st, 2022

It didn’t feel like a journal this year, it felt like a means of collecting data. I knew that the activities I did over an hour couldn’t be represented in a single category, meaning I needed to expand the amount of time that I was measuring. I decided that 20-minute increments were easily divisible and not nitpicky to the extent that it would be overkill. 

I had rated my previous day, in about as subjective a manner as I could. It was a sliding scale, reflecting on the day as a whole. I knew this year that 24 hours couldn’t be summarized in a single number, so I decided instead to make this something I tracked in 20-minute intervals as well. My original intention with tracking all of my time was to hold myself more accountable. I wasn’t sure how to do this without listing productivity as another variable to track. At this point, I was tracking two more variables over 72 intervals throughout the day instead of 24. This would give me a whole lot more data to work off of, but it could lead to getting too specific, on the other end of the spectrum than the previous year.

January 12th, 2022
Binary Analysis

I spent the night meditating before I went to bed. I slept for six hours (not tracking that in a more overt manner) before I spent several hours in the morning working on bracketology before I ate and cleaned for a little bit. I did some reading after lunch before I showered, had supper, and walked over to my girlfriend’s apartment. We talked for several hours, it was the first time I had seen her since Christmas Break, and we watched a movie in bed together. It was a less productive day than usual, but one where I have a better idea of the specifics than I would a year earlier. 

I knew that not every variable was to be nonbinary and that through tracking specific events, I would need to consider binary occurrences in my life throughout the day. Events I wanted to do less of, or at the very least make myself more aware of. I showered during the night, not the day on that Wednesday in January, had sex, socialized, wrote in my journal, and slept in my bed. I spent more time in that bed in the morning than I would’ve wanted, the bed that I only got six hours of sleep in the previous night. I didn’t spend money, go on a substantive enough walk, do drugs, or make a hard choice. I deemed it not a productive day, a variable existing independently of my productivity I tracked every 20 minutes.

I didn’t realize that I was building a multivariable linear equation of my productivity and mood at the time, but that's what this was progressing towards.

June 7th, 2022
Early-June Audible

Of the objective variables I knew I could track, one was quite obvious. The issue with starting at the beginning of the year was that I felt as if revisions had to wait until the start of the next year. I instead decided on this day, after sleeping at my girlfriend’s place, coming home, walking to the store, then biking to the library that I would spend the next six hours at the library retroactively tracking my location for the previous six months, so to have it as something I used moving forward. I began tracking the standard deviation of my location, productivity, and mood.

June 30th, 2022

In having this much data with this many variables, there is room to have eye-appealing graphs, the mid-year point giving me the ability to reflect upon a good amount of data. I had data in categories and subcategories. The majority of my time was spent socializing and sleeping. Having graduated a month earlier in May, I spent twice as much time on basketball than I did studying, even including my work on job applications. 

September 6th, 2022
A Necessary Adjustment

I started working a job in July, something I hadn’t mapped out the variables for in January. There are a finite number of colors, especially with having split these into subcategories. I added in a lot more green after starting at an 8-4, causing a minor headache with remaking the graphs, but it was a problem that was easier to solve than I had previously considered, allowing me to not present my possible actions for the year in January.

December 31st, 2022
35,000 Decisions a Day…

Though mood and productivity were the variables I held the most interest in, running my dummies against sleep as the dependent variable was interesting to see as well. 

Unsurprisingly, the seven hours of sleep were the most impactful. Secondly was whether or not I filled out a Day Log, being strongly negatively correlated. This isn’t to say journaling made me lose sleep, but the fact that I stopped doing the day log once I had gotten a job. I’m sure if I graphed my sleep against the time, I would see that I slept a lot more in the second half of the year, while I had an 8-4, rather than when I was a student, on a more erratic schedule.

The line of best fit shows this in a way the graph does not as well. At the very least, my sleep schedule was a lot less erratic at the end of the year than it was at the beginning. Drugs are positively correlated with the amount of sleep I got, context is necessary for this. It is important to remember I counted sleep from the start of the day, meaning if I drank on a Saturday night, the sleep would be counted from the Friday night/Saturday morning. Drinking didn’t mean I’d sleep more, but likely the fact I was on a weekend and could sleep later.

The highest correlated dummy variable to my mood was the productive dummy (the binary variable, not the one I tracked hourly). Though it is not significant, money seemed to buy happiness in 2022, likely more tied to the fact that I was more active on days I was spending money rather than spending the day inside. 

The greatest correlation was found with productivity as a dependent variable. Admittedly, some of my variables are automated, specifically sleeping and walking. Productivity is on a 1-5 scale, and sleeping is set to 3, while walking is set to 4. Even if it weren’t automated, the fact that I was more productive on days when I got outside makes sense. On the other end of the spectrum, days, where I spend too much time in my bed after waking up, leading to less productivity, make sense. 

The socialization variable should be all but discounted. At the beginning of the year, I checked that dummy whenever I saw my girlfriend. She moved in in August, meaning I’ve checked it every day. Though it wasn’t difficult to retroactively check my location, the subjective nature of socialization made this a variable I knew I needed to update in 2023. 

There seemed to be a very slight correlation between productivity and sleep, but the amount of sleep I had didn’t have all too much of a variance in my mood, shown by the overall variance of the size of the data points. 

Here is the final average, and how it looks graphed against the date.

The number of lazy mornings I had stagnated in the fall, rising once again around the holiday season once I got more time off of work and went back home. 

Data Presentation Issues

How the nonbinary data was collected made it difficult to do an analysis based on an entire day. Excel isn’t smart enough to only select the only fourth row and collect all 72 data points on that day. Given that there were over 105,000 data points, this would also be difficult to compile manually. Thus, I wasn’t able to do something based on the time of the day, instead needing to compile aggregates of each day. I ran a COUNTIF for each variable, for each day, then I cleaned up the data. This led to slightly less ornate and sophisticated data, but it still allowed me to analyze mood and productivity against my daily, 20-minute actions. 

Sleep was not heavily correlated to very much, but the most significant was related to my job. On days when I researched fraud, essentially every day at work, I sleep better. This makes sense, as it was likely a weekday. 

Watching college basketball was positively correlated with sleep. While the NBA was negatively correlated, though not significant to the extent where takeaways could be made. 

As for mood, there are more correlative factors and subsequent interesting takeaways. As a sort of comedic anecdote, having a lecture to attend negatively impacted my mood for that day. 

The strength of this correlation (p-value of .0000265) meant that running a regression on the day of the week against these factors would be a necessary analysis. Meditation has a negative coefficient and a strong significance (significant at p < 0.01). I find myself more likely to reflect and sit down and think in times of stress and unhappiness. 

On days where I likely find myself sitting inside for an extended amount of time and watching basketball all day, I wasn’t as happy on days where I seemed to get outside. This is based on the negative coefficient for College Basketball, as well as the positive coefficients for when I was either outside in some form.

Though it wasn’t shown in the previous regression, productivity is shown to have a very strong positive correlation with my mood.

This isn’t demonstrated well in graphical form and is shown better through the regression.

Subjectivity

When looking towards the regression on productivity, the subjective nature of it was found to be an issue, to the extent that drawing takeaways might be difficult.

Watching basketball was shown to be strongly negatively correlated with productivity, as something I’d have on in the background, or just while sitting on the couch. The autocorrelation nature of some of the variables made any analysis not worthwhile. 

At this point, I began asking myself the reason for this analysis. Was I doing this to make myself more productive? Or was this just to know and track my actions? The second seemed more likely. If I was looking to have more productive days, I run into two issues with how subjective this process is. Firstly, I wouldn’t likely enjoy what I was doing. Secondly, due to how subjective this whole process is, I might juice the numbers to favor actions that I would prefer. I’d rather watch NBA Draft film than be in a meeting for work, but one is shown to be more productive than the other. 

Even if my mood was better for the more productive activities, this doesn’t mean that action should be preferred over the other. The biggest takeaway from this project has been the importance of balance. 

I am not a machine, nor am I programmed to only enjoy specific activities. Through the last two years of tracking my activities, I’ve come to learn and appreciate the naturalness of everyday actions. I’d rather not spend all of my time inside, nor do I wish to spend money daily, but life happens, and what versions of ourselves are we working on if not allowing us to cheat on specific days? What are we building towards, a perfect, idealized version of ourselves that goes on a walk every day, gets seven hours of sleep a night, and actively lessens their usage of time on activities shown to make themselves unhappy? Or are we just the accumulation of all the good and bad, the happy and unhappy, and the productive and the unproductive? I think we should err on the side of the latter.