def eval_sections_independent( values, nsec=10 ): et = ml_tools.eval_tools() a,p,v = values s = np.sqrt(v) inds = np.argsort( s ) chunks = chunk_it( inds, nsec ) fraction = [] err = [] means = [] for i,c in enumerate( chunks ): ca = a[ c ] cp = p[ c ] cs = s[ c ] fraction.append( float(i+1)/nsec ) means.append( np.mean(cs) ) err.append( np.sqrt( et.mean_square_error(ca,cp) ) ) return np.array( fraction ), np.array( means ), np.array( err )
def eval_sections_overall( values, nsec=10 ): et = ml_tools.eval_tools() a,p,v = values s = np.sqrt(v) inds = np.argsort( s ) chunks = chunk_it( inds, nsec ) fraction = [] err = [] means = [] for i in range( len(chunks) ): sub_chunks = chunks[:i+1] selection = np.concatenate( sub_chunks ) sa = a[ selection ] sp = p[ selection ] ss = s[ selection ] fraction.append( float(i+1)/nsec ) means.append( np.mean(ss) ) err.append( np.sqrt( et.mean_square_error(sa,sp) ) ) return np.array( fraction ), np.array( means ), np.array( err )