Clustering em perguntas morais de pesquisa municipal em São Paulo - GPOPAI USP. Segundo semestre de 2019
Visualizing previously clustered data (KMeans). Each point is a person's response of a moral questionnaire that we a applied as a municipal survey in São Paulo. The questionnaire contains 11 questions made by a liberal point of view and 11 by a conservative point of view. On this notebook, I investigate axes and coordinates, more than the whole data file. I begin with isolating the "mean axis" of each group, defined by ingroup cosine distances, and their relative distances, off-group cosine distances. The final result is a heatmap, by which we can see the relative distances between selected group of people, such as political activists, feminists, autodeclared conservatives/liberals, religious people etc. I tried also visualizing the cosine distances in 1 and 2 dimensions, the linear and circular plots. The final results are on the pdf file.