search

murrayds / sci-mobility-emb

Embedding of scientific mobility across institutions, cities, regions, and countries
4 stars 0 forks source link

Revisions newmath #134

Closed murrayds closed 3 years ago

murrayds commented 3 years ago

Add figure address #130 showing the difference in R2 values across distance metrics. Also breaks down performance by definition of organizations' population (all, mobile-only, raw frequency in trajectories), and a subset of mobility (all, domestic only, international only)

image

I also updated the existing hyperparameter plot adding the gamma value. In addition, I worked to make it prettier, and of a consistent style with other plots

image
jisungyoon commented 3 years ago

It looks nice! Minor comment, can you put the * mark next to the best performance for each population term? I think our embedding is highest, but for better visibility? like this one Screen Shot 2021-01-18 at 4 03 49 PM

murrayds commented 3 years ago

@jisungyoon

Here is the updated figure with the top-performing better marked:

image
murrayds commented 3 years ago

In these last few commits, I added a plot showing the RMSE of predictions made with the gravity model using a range of distance metrics, in the same style of the previous R2 figure. This fulfills issue #131

image
jisungyoon commented 3 years ago

Yeah, I think it looks nice. Let's merge this in and start to write a rebuttal letter!