Changes between Initial Version and Version 1 of Ticket #346


Ignore:
Timestamp:
Mar 11, 2008, 7:47:46 PM (13 years ago)
Author:
Peter
Comment:

Legend:

Unmodified
Added
Removed
Modified
  • Ticket #346 – Description

    initial v1  
    1818Implementation-wise there is probably no reason to perform the normalization back and forth. Instead one could calculate the correlation distance using !AveragerPairWeighted and thereby getting the nearest neighbors. the trick is then to calculate the imputation value directly using averages and variances of x and y.
    1919
    20 `y = s_x * x' + m_x = s_x * y' + m_x = s_x * (y-m_y) / s_y + m_x `
     20`y = s_y * y' + m_y = s_y * x' + m_y = s_y * (x-m_x) / s_x + m_y `
    2121
    22 which is simply the equation for LS regression.
     22which is similar to equation for linear regression
     23
     24`y = s_y/s_x * r * (x-m_x) + m_y`.