Data Study on Our Diets
The skills the author demoed here can be learned through taking Data Science with Machine Learning bootcamp with NYC Data Science Academy.
The background story
There are so many online data recommendations on how to leave a healthy life, the diets that people should follow in order to stay healthy and avoid illnesses or diseases. In the face of interminable suggestions, someone may get easily confused on what is healthy and what food may cause severe diseases if consumed over long period of time. Is fat, meat or sugar really bad for you? It is something that would be the root cause for curing the disease instead of possible expensive ineffective treatments? These are the questions that inspired me to pick this topic for my Shiny R project.
The project data set
The cancer dataset is obtained from OECD website. Unfortunately, the data is very comprehensive. The dataset contains cancer rates for countries of the world over the period of 2000 - 2012 calendar years. The data is rather sparse for possible reasons. Cancer patient and statistical data on cancer is private and it is sitting on the medical facilities' hard drives.
The dataset regarding the food consumption is much more available. The dataset used contained information of fat, sugar and vegetable consumption per capita for the respective countries.
The goal is to visualize the frequency and correlation for between the cancer rates and the food consumption rates via side by side maps.
The Shiny application Data
The shiny application's contains multiple tabs to access the graph, map or data related to the data.
On the first tab the user can select the year and the country to observe the cancer rates. Table 1. shows the illustration below.
Table 1. Cancer types and rates for a given year for a country.
In order to compare it with other countries and see the contrast the user may go to maps tab. The first color map shows the cancer rates for the given year for all the countries for which the dataset contained related data. By hovering over the countries with the cursor the values become apparent.
The user then go and see the second color map where the food consumption for the world can be seen for the given year. The user can select among sugar, fat and other food types.
Conclusion and future improvements
The dataset could give a lot more insight if there was historical data available for most of the countries in the world. From the visualization I found interesting similarities on sugar consumption and cancer rate color maps. Does it mean I proved something? Not really. Cancer develops over long period of time and historical data dating back over decades shall be compared with current cancer data rates. Also, in order to see better correlation the plan is to add scatter plots to see trends and correlation between with food consumption and cancer rate for top ten or maybe twenty countries.
About Author
