Individual blog post 3

#

This past week has been an absolute whirlwind between finishing up DS3000, working on the project, and guest speakers. In addition to all of this, our dialogue had the unique opportunity to visit the beautiful city of Luxembourg for the weekend. During this outing, Dr. Gerber and several of the students (including myself) had an absolutely amazing time playing basketball for hours on end. Additionally, 2 of our classmates celebrated their 21st birthday in style at a delicious pizza restaurant! (Who in the right mind doesn’t love melted ice-cream cake??)

This week, I spent a lot of time applying my best efforts to this rapidly evolving project. Because I am the only DS3000 student in my group, I handle most of the data science aspect of this project. For the third phase of this project, it mainly revolved around refining the first ML model for the project.

As a group, we decided to take the project in a different direction than the proof of concept ML model last exemplified in phase 2 was headed in. Because of this, Dr. Gerber helped me to go through the data and view certain trends to visualize how the data might be manipulated to help our group create a new model. We ultimately landed on using age/gender demographics after hot encoding the dataframe to include every country in the columns. After this, the linear regression model was modified to exponential regression based on the slope of the line for asylum applications over time. This appeared challenging, although in the end it was only one or two extra lines of code. I finished up my contributions to phase 3 by viewing scatterplots of the residuals to check for homoscedasticity and Q-Q plots for residual distribution. Unfortunately, the results for both of these were ultimately dissapointing, but the r2 of 0.26 and relatively low mse were enough proof of concept to continue.