fictive data added to data.py

Author: Julien Roussel

examples/1_timeSeries.ipynb

@@ -53,7 +53,6 @@ def get_params(self):
         return dict_params
 
     def get_params_scale(self, D):
-        print(D.size, D.shape)
         mu = D.size / (4.0 * utils.l1_norm(D))
         lam = 1 / np.sqrt(np.max(D.shape))
         dict_params = {"mu": mu, "lam": lam}
@@ -67,9 +66,6 @@ def decompose_rpca(self, D: NDArray) -> Tuple[NDArray, NDArray]:
         mu = params_scale["mu"] if self.mu is None else self.mu
         lam = params_scale["lam"] if self.lam is None else self.lam
         Omega = ~np.isnan(D)
-
-        print("mu:", mu)
-        print("lam:", lam)
 
         D_norm = np.linalg.norm(D, "fro")
 
@@ -126,15 +122,12 @@ def fit_transform(
 
         # U, _, V = np.linalg.svd(M, full_matrices=False, compute_uv=True)
 
-        print("end")
-        print(M.shape)
         if X.shape[0] == 1:
             M = M.reshape(1, -1)[:, :X.size]
             A = A.reshape(1, -1)[:, :X.size]
         M = M.T
         A = A.T
         # return M, A, U, V, errors
-        print(M.shape)
         return M
 
 

qolmat/imputations/rpca/pcp_rpca.py

@@ -48,7 +48,6 @@ def _prepare_data(
         """
         Transform signal to 2D-array in case of 1D-array.
         """
-        print("shape:", X.shape)
         n_rows_X, n_cols_X = X.shape
         if n_rows_X == 1:
             if self.n_rows is None:

qolmat/imputations/rpca/rpca.py

@@ -320,8 +320,6 @@ def fit_transform(
         """
         X = X.copy().T
         D_init = self._prepare_data(X)
-        print("D")
-        print(D_init.shape)
         omega = ~np.isnan(D_init)
         proj_D = utils.impute_nans(D_init, method="median")
 

qolmat/imputations/rpca/temporal_rpca.py

@@ -72,7 +72,7 @@ df["Sales"] = df['Sales'].astype(float)
 cols_to_impute = ["Sales"]
 
 ```python
-download = True
+download = False
 df_data = data.get_data_corrupted(download=download, ratio_masked=.2, mean_size=120 , groups=["station"])
 
 # cols_to_impute = ["TEMP", "PRES", "DEWP", "NO2", "CO", "O3", "WSPM"]
@@ -94,7 +94,7 @@ df0 = df_data
 
 ```python
 # df_data = df0[df0.index.get_level_values("station").isin(["Gucheng"])]
-df_data = df0[df0.index.get_level_values("station").isin(["Gucheng", "Aotizhongxin"])]
+# df_data = df0[df0.index.get_level_values("station").isin(["Gucheng", "Aotizhongxin"])]
 ```
 
 ```python
@@ -124,8 +124,9 @@ df_data = df_data[["TEMP"]]
 ```
 
 ```python
-imputer_rpca = imputers.ImputerRPCA(groups=["station"], method="PCP", columnwise=True, period=365, max_iter=1000)
-# imputer_rpca = imputers.ImputerRPCA(groups=["station"], method="temporal", columnwise=True, n_rows=365, max_iter=1000, tau=1, lam=0.7)
+# imputer_rpca = imputers.ImputerRPCA(groups=["station"], method="PCP", columnwise=True, period=365, max_iter=1000)
+imputer_rpca = imputers.ImputerRPCA(groups=["station"], method="temporal", columnwise=True, max_iter=1000, period=10, tau=2, lam=0.3, list_periods=[10], list_etas=[0.01], norm="L2")
+
 ```
 
 ```python
@@ -141,11 +142,8 @@ df_imputed.iloc[:365 * (df_imputed.size // 365)]
 ```
 
 ```python
-D = df_imputed.iloc[:365 * (df_imputed.size // 365)].values.reshape(-1, 365).T
-```
-
-```python
-plt.plot(D)
+plt.plot(df_data.loc["Wonderland"], ".", color="black")
+plt.plot(df_imputed.loc["Wonderland"])
 ```
 
 This part is devoted to the imputation methods. The idea is to try different algorithms and compare them.

qolmat/notebooks/benchmark.md

@@ -44,7 +44,7 @@ def get_data(datapath: str = "data/", download: Optional[bool] = True):
         return df
     else:
         city = "Wonderland"
-        x = np.linspace(0, 4 * np.pi, 200)
+        x = np.linspace(0, 40 * np.pi, 1000)
         y = 3 + np.sin(x) + np.random.random(len(x)) * 0.2
         datelist = pd.date_range(datetime(2013, 3, 1), periods=len(y)).tolist()
         dataset = pd.DataFrame({"var": y, "datetime": datelist[: len(y)], "station": city})