Heiko is worrie that we will loose it :)

gistfile1.txt

// Immutable features

// linear model
 - dot prod:
    - pairs
    - 
 - cov var matrix: there's a Feature operator => CovarView => matrix
 -
 

class Features {

  Features(){}
  
  ...
  Features transformed_by(Transformer t) const;
  
  // this evaluates the stacked operators over the features
  // and returns the copy of features
  Features cached() const;

protected:
  void add_flag(flag) {
   flags |= flag;
  }
  int get_flags()
}
// lala land :)
class Trait {
   Trait(Features);
   Trait(View);
   
   iterator begin() = 0;
   iterator end() = 0;
}

class DotTrait {
   DotTrait(Features);

   iterator begin();
   iterator end();
}

// YOLO++ :)
class CovarTrait {
  CovarTrait(DotTrait){}
  
  Matrix get () {

  
  }
}

df = DataFrame()
f = Features(df).tranformed_by(Mean).transformed_by(Normalize).cached();

LinearModel {
  LinearModel()
  
  
  train(Features) {
       // assert
       assert(features.get_flags() & CENTERED);
       
       // we do not expose on model level the type of matrix/vector
       // linalg should find it out, we have an opaque Matrix/Vector class
       // which is type agnostic.
       auto covar = CovarTrait(DotTrait(features)).get();
       
       // inplace,... or not :)
       linalg::add_diag(covar, get("lambda"));
       set("w", linalg::cholesky_solve(covar, y));
  
  } 
}

batch optimisation is taken care automagically
model.train(SubsetView(features))