Missing value imputation in python using KNN

Clock Slave picture Clock Slave · Jul 26, 2017 · Viewed 37.3k times · Source

I have a dataset that looks like this

1908    January 5.0 -1.4
1908    February    7.3 1.9
1908    March   6.2 0.3
1908    April   NaN   2.1
1908    May NaN   7.7
1908    June    17.7    8.7
1908    July    NaN   11.0
1908    August  17.5    9.7
1908    September   16.3    8.4
1908    October 14.6    8.0
1908    November    9.6 3.4
1908    December    5.8 NaN
1909    January 5.0 0.1
1909    February    5.5 -0.3
1909    March   5.6 -0.3
1909    April   12.2    3.3
1909    May 14.7    4.8
1909    June    15.0    7.5
1909    July    17.3    10.8
1909    August  18.8    10.7  

I want to replace the NaNs using KNN as the method. I looked up sklearns Imputer class but it supports only mean, median and mode imputation. There is a feature request here but I don't think thats been implemented as of now. Any ideas on how to replace the NaNs from the last two columns using KNN?

Edit: Since I need to run codes on another environment, I don't have the luxury of installing packages. sklearn, pandas, numpy and other standard packages are the only ones I can use.

Answer

Miriam Farber picture Miriam Farber · Jul 26, 2017

fancyimpute package supports such kind of imputation, using the following API:

from fancyimpute import KNN    
# X is the complete data matrix
# X_incomplete has the same values as X except a subset have been replace with NaN

# Use 3 nearest rows which have a feature to fill in each row's missing features
X_filled_knn = KNN(k=3).complete(X_incomplete)

Here are the imputations supported by this package:

•SimpleFill: Replaces missing entries with the mean or median of each column.

•KNN: Nearest neighbor imputations which weights samples using the mean squared difference on features for which two rows both have observed data.

•SoftImpute: Matrix completion by iterative soft thresholding of SVD decompositions. Inspired by the softImpute package for R, which is based on Spectral Regularization Algorithms for Learning Large Incomplete Matrices by Mazumder et. al.

•IterativeSVD: Matrix completion by iterative low-rank SVD decomposition. Should be similar to SVDimpute from Missing value estimation methods for DNA microarrays by Troyanskaya et. al.

•MICE: Reimplementation of Multiple Imputation by Chained Equations.

•MatrixFactorization: Direct factorization of the incomplete matrix into low-rank U and V, with an L1 sparsity penalty on the elements of U and an L2 penalty on the elements of V. Solved by gradient descent.

•NuclearNormMinimization: Simple implementation of Exact Matrix Completion via Convex Optimization by Emmanuel Candes and Benjamin Recht using cvxpy. Too slow for large matrices.

•BiScaler: Iterative estimation of row/column means and standard deviations to get doubly normalized matrix. Not guaranteed to converge but works well in practice. Taken from Matrix Completion and Low-Rank SVD via Fast Alternating Least Squares.