// context. data: presidential elections 2015, http://datar.noip.me/en/dataset/elecciones-ncionales-2015-resultados-provisorios

// the grieta:

// but it's a fallacy. no hegemony:

// # Min. 1st Qu. Median Mean 3rd Qu. Max. // # 0.2409 0.4015 0.4609 0.4853 0.5448 1.0000 // 36.3% > 0.5

// few "FPV poll stations" or "CAMBIEMOS poll stations." if poll stations represent (local) communities, then communities have preferences, but weren't "of" each main party.// you can find communities along any degree of intensity for each big party.

// question: to what degree was then Massa an independent third party?

// independent of FPV ~ adj R^2 ~ 0

// slightly anticorrelated to CAMBIEMOS ~ beta ~ -0.19, adj R^2 ~ 0.13

// what it was, is anticorrelated strongly with the sum of both (-0.72, aR^2 0.66)

// this isn't a priori necessary: doens't happen for IZQ

// question becomes: what communities are there? how do they look?

// one observation is that generally speaking, GEN/IZQ/SAA's level is independent of MASSA, with adj. R^2 of ~0.1 or much less.

None

// clustering isn't neat (wouldn't be expected with so high dimensionality) but it does stabilize ~ 4 (in other words: no strong clusters, but clusters as model)// 4D is kind of messy to look at:

// PCA gives (n-1) parameters, etc