Data Recording

Lifestyle Data Recording

As I explore the recording process, I will share my embarrassing lifestyle captured in April and how I progressed to live a more consistent life in a later month.

Since the purpose of the self-evaluation project is to analyze my lifestyle, recording meaningful data that can capture my day-to-day life is critical. Data is powerful because it provides insights that would otherwise be impossible to find if anyone was to analyze by simply looking back at one's life. It also allows the implementation of advanced Data Science techniques such as machine learning to create forecasts or to get suggestions on decision making. By utilizing Python and Notion API, I wanted to create an evaluation system that can be applied towards creating healthy habits, tracking project progress, or even evaluating the company's growth. In the end, I successfully assembled such a system by utilizing Python and various APIs.

Thus, first, we will begin by explaining which variables best describe my daily performance and how they are recorded.

Notion: Data Collection

To collect daily data, I have made good use of a collaboration platform called Notion. It was a suitable fit for my project since it allowed extensive data accumulation. It also helped handle diverse datasets, make functions within the platform, and connect to Python using its REST API.

Every night, corresponding values are plugged into each variable, shown on the left. Once recorded, created functions instantly calculate them into intended ratios. Note that the values shown below are the "perfect" case scenario, where it represents the most productive and abundant day.

Data Sources & Explanation

Manaul Recording

Application Recording

Automated Recording

Rise time: Time recorded when I got out of bed, where 0 indicates that I woke up on the intended rise time (usually 8 am). A negative value is inputted for rising earlier, and a positive value for waking up later than the intended time.
Screen time & Phone pickups: The iPhone Screen time feature is used to record the values.
Total To-do list & finished: To record the number of to-do lists, the Notion Scheduling database is used. (See Notion Automation page)
Meditation (min): Meditation is done through an application called Balance. The total duration is usually 10, 15, or 20 minutes.
Personal Reading (min): Reading duration is measured by simply timing the total minutes spent reading.
Drink: If I feel 100% sober, then 0 is inputted, whereas if I feel the alcohol even a slightest, 1 is recorded.
Books finished : 0 if I have not completed any book on a given day and 1 if I have.
Run (km): Adidas application calculates the total distance ran in km.
Star Metrics: These five different variables collaboratively depict the fullness of a given day. I believe these metrics are the best way to adequately represent a subjective view of the day-to-day rating. The average of these values is calculated and applied to compute the ratio on the right side.

Input Variables

This image above illustrates different variables that impact my daily life. They were chosen through numerous experiments and lessons learned by reading books and enrolling in courses taught by world-class leaders. Hence, this table consists of 16 unique variables that are meaningful indicators that could represent the abundance of each day.

The variables include the time I woke up, minutes spent on my phone, total to-do lists completed, minutes spent meditating, and others. They are inputted every single day and currently have over 2 years worth of evaluations. Also, to prevent inconsistency, I have tried to minimize the "missed days" as much as possible.

Output Variables (%)

Depending on my preferred ratio and the numerical type of each variable, different Mathematical models are applied.
(Notion code example below)

Notion provides a built-in function where basic coding can be implemented within the dataset. As shown above, these mathematical functions generate unique outputs(%) to demonstrate how I lived each day.

Notion: Built-in Coding feature

Total to-do list (%)

The above shows that when there are less than 14 tasks, it calculates the percentage to have less ratio. In other words, the more tasks I have each day(more than 14), the more it would affect the total percentage.

Rise Time (%)

For the convenience of data handling, I have set the default wake-up time to 8:00 am. Thus, the input value 0 demonstrates waking on time at 8 am. A positive value would indicate oversleep, and a negative value would mean waking up earlier than 8.

How are the recorded data visualized?

The top plot is from the data recorded in April 2021, and the bottom is from November 2021. These two visualizations best represent the different lifestyles of different time periods. As shown below, there were numerous fluctuations and inconsistencies in April. On the other hand, the plot of November is more stable and fulfilling. With these data visualizations, different lifestyles and motivation levels can be easily observed.

Plot of April, 2021

Top Plot Description

Purple line: total percentage of how I lived each day (left y-axis)
Green dashed line: wake-up time (right y-axis)
Red bar: Reading (minutes)
Yellow bar: Meditation (minutes)
Red dot: Days I drank beer

Bottom Plot Description

Rise Time Histogram: It gives average wake up time and its distribution

Plot of November, 2021

Observations

The further apart the purple line is from the green dashed line, the more likely it was a productive day.
Total average in November (72.73%) was significantly higher than it was in April (47.25%).
Right skewed Rise Time Histogram indicates that there were more early risen days, which was the case for November.

Connecting to Notion API

If these datasets were to be downloaded every time new evaluations are recorded, that would be extremely redundant. Hence, writing a code that directly connects to the Notion API enhances efficiency when retrieving data.

With such data, there are powerful things that can be done. In my case, these are a few analyzations that were done through python and R.

Visual Analyzation (using Matplotlib Module)
Various correlation tests among variables
Automate visualization update using Python
Connect to Slack API for hourly status update
Linear Regression Model
Time series Analysis (In Progress)