American Health: Exploration of the Obesity Epidemic

Posted on Jul 29, 2018

The skills the authors demonstrated here can be learned through taking Data Science with Machine Learning bootcamp with NYC Data Science Academy.


America is consistently ranked as one of the top 20 most obese countries in the world. Obesity is not only a major health issue but can lead to numerous diseases and increased risk of heart disease, diabetes, sleep apnea, as well as many other health problems. Understanding the cause of obesity and  demographics associated with the epidemic could be helpful to mitigating the problem overall.


The data I used for this study came from the US Census in 2014. It was specifically made up of questions asked by the US Bureau of Labor Statistics. There were 37 variables related to height, weight, income, eating behavior, drinking behavior (non- alcoholic beverages), and exercise behavior. The variables analyzed in this study were body mass index (BMI, which is defined as meters^2/weight in kilograms), income, exercise frequency, and eating out frequency. Data Source: Link to data.

Exploratory Data Analysis:

Before diving into my questions about the data, I selected variables that could have an interesting relationship with BMI. Examining the distribution of my selected variables helped me see the levels of the variables, the missingness in each variable and where most observations lied.

Figure 1: Distribution of BMI for Americans in 2014

5% of BMI was missing. Most Americans seem to be either overweight or obese. This confirmed my research that American obesity is in fact prevalent.

How does BMI vary with income?

The data included a variable for income where the individual checked if their income was above, below or equal to 185% of the poverty threshold (abbreviated as PT in the following graphs). There were some missing groups like if the individual left the question blank, refused to answer the question or didn't know their income. I examined how this variable interacts with BMI.

Figure 2: BMI distribution by Income

As figure 2 demonstrate, when an individual's income is lower, they are more likely to be overweight or obese. Individuals with higher incomes tend to be either normal weight or overweight. This discrepancy could be because unhealthy food tends to be much more affordable. Some of the most affordable meals come from fast food joints which are often loaded with toxic chemicals that are detrimental to human metabolism. Similarly, non-organic food tends to be cheaper and much more harmful than the pricier, organic alternative. It makes sense that those with a lower income would turn to these cheaper food choices and have a higher BMI as a result. A chi-squared test on these two variables revealed that BMI and income group are not independent of one another.

I also examined the distribution of missing BMI by income group to determine if the distribution was uniform and if missing BMI could be excluded from the study.

Figure 3: Missing BMI

Figure 3 shows that most individuals who did not report their BMI were from a high income group. Excluding the missing BMI observations in this study could introduce bias. Perhaps a better solution would be to impute values for the missing BMI based on income group rather than excluding them all together. A chi-squared test of the missing BMI versus income group similarly revealed that these two groups are not independent.

Last, I looked at the missing income groups to see if I could draw any conclusions about their BMI based on the distribution of BMI.

Figure 4: Missing Income Groups

Figure 4 shows that those who refused to answer what their income was were mostly normal weight or overweight. This distribution is similar to the income group of those with an income greater than 185% of the poverty threshold. The distributions of those who did not know their income or left the question blank is less conclusive. Imputing values for these groups may require a deeper dive into other variables.

Does exercise frequency vary by income group?

After concluding that high earners tend to have a lower BMI, I looked into whether higher earners are more likely to exercise as well.

Figure 5: Exercise Frequency by Income

Clearly, most high earners exercised in the past 7 days. Only about 50% of lower earners exercised in the past 7 days. This could be because high earners have more access to gyms, personal trainers and in general are surrounded by other high earners who prioritize exercise. Lower earners may not have access to these facilities.

I looked at the missing income groups as well here to compare the distribution of the missing groups to figure 5.

Figure 5: Missing Income vs Exercise

More people who responded saying they don't know their income or refused to answer what their income was were more likely to have exercised in the past 7 days. Both of these distributions seem more similar to the high earners. This confirmed my suspicions that those who refused to answer what their income was were more likely high earners from figure 4.

Is the mean BMI for those who ate out in the last 7 days different from the mean BMI for those who did not eat out in the last 7 days?

Last, I examined how eating behavior could influence BMI by comparing the distribution of BMI for those who typically eat out compared to those who typically cook their meals.

Figure 6: BMI distribution by Eating Behavior

The spike where BMI = 0 is for the 5% of data where BMI was missing. Aside from that,the distribution of BMI for both of of these groups is normal as demonstrated by the graph. Also the mean and median of each distribution is approximately equal which confirms normality. This makes sense because of course most people would have a BMI around the same ballpark.

The mean BMI of those who eat out was 26.8. The mean BMI of those who do not eat out is 25.6.
I performed a t-test to test if the difference in the BMI mean of those who eat out and the BMI mean of those who do not eat out was statistically significant  The t-test revealed that I could reject the null hypothesis that the mean of these two groups are equal.

Eating out frequently does in fact result in a higher BMI. This makes sense because food that is not home cooked tends to be saturated with oils, lower in nutrients and less likely to be organic.


The conclusions I can draw from this study is this:

  • Higher earners tend to have a more normal BMI which is lower than lower earners
  • Higher earners are more likely to exercise
  • Those who eat out frequently tend to have a higher BMI

Being able to confirm these conclusions with statistical tests allows us to better understand American obesity and understand how to solve the problem. Some solutions would be:

  • Make healthy food more affordable for lower earners
  • Make exercise facilities more accessible and affordable for lower earners
  • Impose regulations that mandate restaurants to serve mostly healthy options, or discourage society from eating out frequently


It's important to recognize that this data came from self reported US Census surveys. Self reported data can be less trust worthy because individuals may be less inclined to be honest.

Also, while body mass index is a generally accepted metric of health, it would be interesting to consider other health metrics if they were available. Body mass index tends to ignore the fact that muscle weighs more than fat, so an individual can appear "overweight" or "obese" by BMI standards but have a low body fat percentage and be in great health. If it were available to me, I would have liked to also consider body fat percentage, bone density and waist circumference.

Link to app:

Link to Code:

About Author

M Sunanda

Sunanda is a data science fellow at NYCDSA and an Associate of the Casualty Actuarial Society. She worked as an actuary for 6 years before joining NYDSA. Sunanda enjoys the storytelling and sheer power capabilities of data science....
View all posts by M Sunanda >

Leave a Comment

No comments found.

View Posts by Categories

Our Recent Popular Posts

View Posts by Tags

#python #trainwithnycdsa 2019 2020 Revenue 3-points agriculture air quality airbnb airline alcohol Alex Baransky algorithm alumni Alumni Interview Alumni Reviews Alumni Spotlight alumni story Alumnus ames dataset ames housing dataset apartment rent API Application artist aws bank loans beautiful soup Best Bootcamp Best Data Science 2019 Best Data Science Bootcamp Best Data Science Bootcamp 2020 Best Ranked Big Data Book Launch Book-Signing bootcamp Bootcamp Alumni Bootcamp Prep boston safety Bundles cake recipe California Cancer Research capstone car price Career Career Day citibike classic cars classpass clustering Coding Course Demo Course Report covid 19 credit credit card crime frequency crops D3.js data data analysis Data Analyst data analytics data for tripadvisor reviews data science Data Science Academy Data Science Bootcamp Data science jobs Data Science Reviews Data Scientist Data Scientist Jobs data visualization database Deep Learning Demo Day Discount disney dplyr drug data e-commerce economy employee employee burnout employer networking environment feature engineering Finance Financial Data Science fitness studio Flask flight delay gbm Get Hired ggplot2 googleVis H20 Hadoop hallmark holiday movie happiness healthcare frauds higgs boson Hiring hiring partner events Hiring Partners hotels housing housing data housing predictions housing price hy-vee Income Industry Experts Injuries Instructor Blog Instructor Interview insurance italki Job Job Placement Jobs Jon Krohn JP Morgan Chase Kaggle Kickstarter las vegas airport lasso regression Lead Data Scienctist Lead Data Scientist leaflet league linear regression Logistic Regression machine learning Maps market matplotlib Medical Research Meet the team meetup methal health miami beach movie music Napoli NBA netflix Networking neural network Neural networks New Courses NHL nlp NYC NYC Data Science nyc data science academy NYC Open Data nyc property NYCDSA NYCDSA Alumni Online Online Bootcamp Online Training Open Data painter pandas Part-time performance phoenix pollutants Portfolio Development precision measurement prediction Prework Programming public safety PwC python Python Data Analysis python machine learning python scrapy python web scraping python webscraping Python Workshop R R Data Analysis R language R Programming R Shiny r studio R Visualization R Workshop R-bloggers random forest Ranking recommendation recommendation system regression Remote remote data science bootcamp Scrapy scrapy visualization seaborn seafood type Selenium sentiment analysis sentiment classification Shiny Shiny Dashboard Spark Special Special Summer Sports statistics streaming Student Interview Student Showcase SVM Switchup Tableau teachers team team performance TensorFlow Testimonial tf-idf Top Data Science Bootcamp Top manufacturing companies Transfers tweets twitter videos visualization wallstreet wallstreetbets web scraping Weekend Course What to expect whiskey whiskeyadvocate wildfire word cloud word2vec XGBoost yelp youtube trending ZORI