big data in insurance
Jonathan Buchanan's Winter 2018 Independent Study
Today I planned to explore the data sets that are publicly available on cms.gov. I know that the amount of data available is almost overwhelming, so I planned to pick one data set and focus on that for the day so I didn't distract myself with all the other data. I wanted to do some brief analysis with it so I could begin to grasp some possible connections, but I also knew I had a limited amount of time today because we had a 2-hour delay before school.
To begin, I began browsing the publicly available data on cms.gov. I had some difficulty narrowing it down to one dataset that interested me, but ultimately I chose to download a table that dealt with the prescription of opiates on a state/county basis. This intrigued me because opiates are an issue that many Americans are affected by, and it would be interesting to gain some insight into that statistics of it. To begin, I plotted the data using Tableau on a map of the United States. You can view the plot in the slideshow at the top of this post. One thing that jumped out to me was that less-densely populated states such as Utah, Idaho, and Nevada had a higher rate of prescriptions that densely populated states such as Texas, California, and New York. To put this idea into statistical terms, I downloaded data from the 2010 US Census, and plotted the population density of each state against its prescription rate of opioids. This graph is in the slideshow as well. In the graph, you can see a trend that, generally speaking, as a state has a greater population density, it has less prescriptions of opiates. To confirm this trend, I added a trend line to the plot. The p-value of this trend is less than 0.0001, which indicates that this is most likely a statistically significant finding. However, the R-squared value is around 30%, which indicates that there is a lot of variation in the data. However, opiate prescriptions are a complex quantity to measure; population density is only one factor. Overall, I was pleased with the results I had come up with already. Overall, I felt as though I were productive today because I picked an issue that was interesting to me and presented some interesting statistics. I am looking forward to digging deeper into this issue. I think my next step is to confirm this relationship between population density and prescriptions on a county basis, which should yield more accurate results do to the increased sample size as well as population density at the county level being a better representation of an area, due to their smaller size. Tomorrow is going to be my first day at Paramount, so I'm looking forward to spending time with the people who do this kind of analysis for a living!
2 Comments
Hannah Spengler
1/10/2018 05:36:54 pm
I LOVE love love that you selected a topic that is meaningful to you. That is awesome. Clearly opioid use is a hot topic right now and something that needs to be addressed and studied. My question to you is...now that you have found statistically compelling evidence that there is a correlation between low population density and opioid prescription, why do you think that is? What understanding can we start to gain from this information? Are there other variables that need to be considered? I am excited to see how you analyze and apply your findings. Maybe you can share these with some of your current colleagues and see what they think, too. Again, great work. I am really enjoying reading about your experience!
Reply
1/12/2018 06:22:06 am
When I made this conclusion before I did the same analysis on the data at the county level which appeared to show no relationship, I thought it was because people living in less densely populated areas tend to be less social, and less social people have a smaller network of people available to help them through a problem like an opioid addiction. However, that appears to not be the case, so we can conclude instead that people living in less dense areas are no more likely to be prescribed opioids. I think other factors that could possibly have a relationship to the prescription rate include medical spending, growth rate of medical spending, and marketing by pharmaceutical companies.
Reply
Leave a Reply. |
ArchivesCategories |