ihodes / fancyimpute

Matrix completion using nuclear norm minimization (naive implementation of Candes & Recht 2008)

fancyimpute

Simple implementation of Exact Matrix Completion via Convex Optimization by Emmanuel Candes and Benjamin Recht using cvxpy.

Usage

from fancyimpute import ConvexSolver

solver = ConvexSolver(
    min_value=0.0,
    max_value=1.0,
    error_tolerance=0.0005)

# X_incomplete has missing data which is represented with NaN values
X_filled = solver.complete(X_incomplete)