Per Capita Income
The Mayor has set a primary goal of increasing the per capita income of Tulsa. To select policies necessary to meet this goal, the city needs to understand the factors correlated to per capita income. Using Census Bureau Data, this report performed feature selection; identifying the factors correlated to income. Results showed the most impactful factors were related to education level, high school drop-outs having the most impact on a tract’s income. For this reason, Tulsa Data Science, Inc, formally recommends that the City work with school districts to increase the high-school graduation rates. The city of Tulsa should perform more research to understand the causality of decreasing high-school drop-outs.
For the first year in office, Mayor Bynum has set the goal of increasing per capita income in Tulsa. Succeeding in this goal, will translate into a real tangible effect of increasing not only income, but the average level of education. In addition, Tulsa will be a more attractive destination for the well-educated, allowing the city to diversify industries and jobs. Finally, by increasing income levels across low income tracts, the city would reduce income disparity across the city.
While improving income can provide benefits, identifying the right policies is a complicated process. This is because a variety of factors influence income level. It is unclear which factors have the most impact. For example, education level, availability of natural resources, and demographics can all have an impact on income. With limited resources, the Mayor’s office must explore technique to understand which factors have the most impact. This will allow Tulsa to make decisions which will improve the long-term income and diversification strategy.
A few techniques were proposed to identify factors which influence income level.
Census bureau data was used to examine the effect of -over a hundred- different income influencers. Using several machine learning algorithms, feature selection was used to find the highest independent variables correlation to income. This allowed the building of a model that more accurately predicted the income level of each of the city’s tracts. Using this approach, it was hypothesized that both education level and social factors would have the most impact on income levels.