Data Analysis on Stress Causes and Outcomes
The skills I demoed here can be learned through taking Data Science with Machine Learning bootcamp with NYC Data Science Academy.
Objective
What does the common data say is the sources of stress and how to manage them for a purposeful impact?
Introduction
What is stress?
Simply put, Stress is the feeling of being overwhelmed or unable to cope with mental or emotional pressure. In the past few years there has been a staggering increase in number of stress related issues. stress results in “accidents, absenteeism, employee turnover, diminished productivity, and direct medical, legal, and insurance costs” that cost the United States $300 billion every year.
According to The American Institute of Stress:
- About 33 percent of people report feeling extreme stress
- 77 percent of people experience stress that affects their physical health
- 73 percent of people have stress that impacts their mental health
- 48 percent of people have trouble sleeping because of stress
Stress can affect all aspects of life, including emotions, behaviors, thinking ability, physical health, professional achevements and personal relationships.
Why do we care?
The first step in fixing any issue is to know the causes. The benefit of this analysis can be two folds. On a personal level Identifying the triggers of stress can help improve lifestyle, emotional and mental health. It can direct towards making the necessary changes to manage and lower stress levels. From a business perspective, focusing on reducing stress can help with increased productivity, energy levels, employee engagement indirectly impacting company’s bottom-line.
There are several organizations that could benefit from stress management exercises by tailoring their marketing programs for a higher purpose and business outcomes. Fitness centers, wellness facilities, meditation and yoga camps, sports wearable companies, food catering and cafeterias.
The Data
For the analysis I choose the lifestyle and wellbeing data from Kaggle.com. This dataset contains the survey responses from www.Authentic-Happiness.com.
There are 24 attributes describing how we live our lives and thrive both professionaly and personally: it reflects how well we shape our lifestyle, habits and behaviors to maximize overall life satisfaction along the following five dimensions:
Healthy body, reflecting your fitness and healthy habits. Healthy mind, indicating how well you embrace positive emotions; Expertise, measuring the ability to grow your expertise and achieve something unique; Connection, assessing the strength of your social network and your inclination to discover the world. Meaning, evaluating your compassion, generosity and how much 'you are living the life of your dream'.
Python libraries used
- Pandas
- Matplotlib
- Seaborn
- Numpy
Exploratory Data Analysis
For this dataset stress is measured on a scale of 0 to 5. Most of the variables are either binary or have fixed set of values. Based on the representation of values I calculated average stress to be the dependant variable. I then analyzed the impact of other variables on average stress levels.
Data on BMI, Fruits and veggie to average stress.
The BMI range in the data is categorized in two groups. Population with BMI Below 25 and above 25. Looking at the graph we can say that stress levels rise with the increase in bmi. People with BMI above 25 have 8.6% increase in average stress levels than people below 25.
1 – BMI below 25 2 - BMI above 25
8.6% increase in average stress
As with our physical health, stress levels are also impacted by the intake of fruits and vegetable, and sure enough the graph reflects the same. People having 5 servings of fruits and veggies in a day experience 16.1 % decrease in average stress.
16.1% decrease in average stress
Data on Sleep hours and average stress
We cannot undermine the importance of sleep for a better quality of life. Very obvious question that pops up looking at the graph is, is it possible for people sleeping just one hour a day to have lower stress levels? Looking into details I found that it comprises of a very small percentage of the survey population, 0.1 % to be precise. I found the number insignificant enough to exclude from the analysis. The resulting graph makes much more sense. There is a decline in average stress with increase in sleep hours.
Social network, core circle and average stress Data
According to the survey question social network represents the number of interactions in a day . One may expect social networking to have a lowering impact on stress levels. but this was an interesting find. The responses range from a value of 1 to 10 in the data. Notice that having no interaction or meeting 10 people may cause stress levels to rise comparatively. On the other hand having a higher number in core circle which counts as number of people close to you can have a declining impact on average stress
Places visited and sufficient income
The number of places visited in a year has positive impact on stress levels
People who visit 10 places in a year experience 21.5 % less stress than people who don’t visit any.
| 21.5% decrease in average stress |
Again income in this dataset is categorized into groups, population having suffient income or hardly sufficient income. people with insufficient income have 16.7% increase in average stress levels.
1 – Hardly Sufficient 2 - Sufficient
16.7% increase in average stress
Weekly meditation and time for passion
Weekly meditation and time for passion have a therupatic impact on stress levels. Based on the given data people who meditate 10 hours a week have 32.1% decrease in average stress levels. While many people may not have 10 hours in a week for meditatation but looking at the graph we can say that even spending 1 to 3 hours may help lower stress significantly.
32.1% decrease in average stress
As per the survey questions, time for passion is described as the total number of hours people spend in a day doing something they like and are passionate about. Spending 1 to 3 hours in a day may help lower stress levels though anytime spend more than 4 hours would not make much difference.
26.2% decrease in average stress
And finally I created a heatmap, to help identify the highly coorelated variables. Each square shows the correlation between the variables on each axis, the larger the number and darker the color the higher the correlation between the two. A positive or a negative number indicates a positive and negative correlation respectively.
Conclusion
Upon analysis over 10 attributes in various dimensions we can conclude the following:
-
- Based on the given data BMI has the strongest positive correlation with stress (coefficient = 0.8)
- Weekly meditation and time for passion have the strongest negative correlation ( coefficient= 0.21, 0.15) respectively.
-
- Balance is the key when it comes to social networking, having no social interaction or having too much may cause stress levels to increase.
- Hours of sleep, exercise and good nutrition can help lower stress reasonably.
- Correlation of stress with people supporting others or donating is not very significant.
About Author
