Merge pull request #33 from Quantmetry/patch_rpca_reshaping

Patch rpca reshaping

Author: JulienRoussel77

.bumpversion.cfg

@@ -1,5 +1,5 @@
 [bumpversion]
-current_version = 0.0.10
+current_version = 0.0.11
 commit = True
 tag = True
 

.gitignore

@@ -57,7 +57,7 @@ qolmat/.converge
 examples/*.ipynb
 examples/figures/*
 examples/data/*
-examples/*.ipynb
+examples/local
 
 
 # VSCode

docs/conf.py

@@ -22,7 +22,7 @@
 author = "Quantmetry"
 
 # The full version, including alpha/beta/rc tags
-version = "0.0.10"
+version = "0.0.11"
 release = version
 
 

examples/benchmark.md

@@ -131,8 +131,8 @@ imputer_spline = imputers.ImputerInterpolation(groups=["station"], method="splin
 imputer_shuffle = imputers.ImputerShuffle(groups=["station"])
 imputer_residuals = imputers.ImputerResiduals(groups=["station"], period=7, model_tsa="additive", extrapolate_trend="freq", method_interpolation="linear")
 
-imputer_rpca = imputers.ImputerRPCA(groups=["station"], columnwise=True, period=365, max_iter=200, tau=2, lam=.3)
-imputer_rpca_opti = imputers.ImputerRPCA(groups=["station"], columnwise=True, period=365, max_iter=100)
+imputer_rpca = imputers.ImputerRPCA(groups=["station"], columnwise=True, period=7, max_iter=200, tau=2, lam=.3)
+imputer_rpca_opti = imputers.ImputerRPCA(groups=["station"], columnwise=True, period=7, max_iter=100)
 
 imputer_ou = imputers.ImputerEM(groups=["station"], model="multinormal", method="sample", max_iter_em=34, n_iter_ou=15, dt=1e-3)
 imputer_tsou = imputers.ImputerEM(groups=["station"], model="VAR1", method="sample", max_iter_em=34, n_iter_ou=15, dt=1e-3)
@@ -154,8 +154,8 @@ dict_imputers = {
     # "OU": imputer_ou,
     # "TSOU": imputer_tsou,
     # "TSMLE": imputer_tsmle,
-    # "RPCA": imputer_rpca,
-    # "RPCA_opti": imputer_rpca_opti,
+    "RPCA": imputer_rpca,
+    "RPCA_opti": imputer_rpca_opti,
     # "locf": imputer_locf,
     # "nocb": imputer_nocb,
     # "knn": imputer_knn,
@@ -203,7 +203,7 @@ results
 
 ```python
 df_plot = results.loc["energy", "All"]
-plt.bar(df_plot.index, df_plot, color=tab10(0))
+plt.barh(df_plot.index, df_plot, color=tab10(0))
 plt.show()
 ```
 

qolmat/_version.py

@@ -1 +1 @@
-__version__ = "0.0.10"
+__version__ = "0.0.11"

qolmat/benchmark/cross_validation.py

@@ -1,9 +1,10 @@
 import logging
 from typing import Any, Callable, Dict, List, Union
-from skopt.space import Categorical, Dimension, Integer, Real
+
 import numpy as np
 import pandas as pd
 import skopt
+from skopt.space import Categorical, Dimension, Integer, Real
 
 from qolmat.benchmark.missing_patterns import _HoleGenerator
 

qolmat/imputations/rpca/rpca.py

@@ -58,3 +58,22 @@ def _prepare_data(self, X: NDArray) -> NDArray:
                 return X.copy()
             else:
                 raise ValueError("`n_rows` should not be specified when imputing 2D data.")
+
+    def get_shape_original(self, M: NDArray, X: NDArray) -> NDArray:
+        """Shapes an output matrix from the RPCA algorithm into the original shape.
+
+        Parameters
+        ----------
+        M : NDArray
+            Matrix to reshape
+        X : NDArray
+            Matrix of the desired shape
+
+        Returns
+        -------
+        NDArray
+            Reshaped matrix
+        """
+        size = X.size
+        M_flat = M.flatten()[:size]
+        return M_flat.reshape(X.shape)

qolmat/imputations/rpca/rpca_noisy.py

@@ -366,7 +366,7 @@ def decompose_rpca_signal(
         elif self.norm == "L2":
             M, A, U, V, errors = self.decompose_rpca_L2(D_proj, Omega, lam, tau, rank)
 
-        M = M.reshape(X.shape)
-        A = A.reshape(X.shape)
+        M_final = self.get_shape_original(M, X)
+        A_final = self.get_shape_original(A, X)
 
-        return M, A
+        return M_final, A_final

qolmat/imputations/rpca/rpca_pcp.py

@@ -99,12 +99,6 @@ def decompose_rpca_signal(
         D = self._prepare_data(X)
         M, A = self.decompose_rpca(D)
 
-        # U, _, V = np.linalg.svd(M, full_matrices=False, compute_uv=True)
-
-        # if X.shape[0] == 1:
-        # M = M.reshape(1, -1)[:, : X.size]
-        # M = M.reshape(X)
-        # A = A.reshape(1, -1)[:, : X.size]
-        M = M.reshape(X.shape)
-        A = A.reshape(X.shape)
-        return M, A
+        M_final = self.get_shape_original(M, X)
+        A_final = self.get_shape_original(A, X)
+        return M_final, A_final

setup.py

@@ -3,7 +3,7 @@
 from setuptools import find_packages, setup
 
 DISTNAME = "qolmat"
-VERSION = "0.0.10"
+VERSION = "0.0.11"
 DESCRIPTION = "Tools to impute"
 LONG_DESCRIPTION_CONTENT_TYPE = "text/x-rst"
 with codecs.open("README.rst", encoding="utf-8-sig") as f: