Merge branch 'dev' into from-conda-to-poetry

Author: Hông-Lan Botterman

examples/RPCA.md

@@ -34,6 +34,10 @@ from qolmat.imputations.rpca import rpca_utils
 from qolmat.utils.data import generate_artificial_ts
 ```
 
+```python
+from qolmat.imputations.imputers import ImputerRpcaNoisy, ImputerRpcaPcp
+```
+
 **Generate synthetic data**
 
 ```python tags=[]
@@ -46,16 +50,33 @@ amp_noise = 0.1
 X_true, A_true, E_true = generate_artificial_ts(n_samples, periods, amp_anomalies, ratio_anomalies, amp_noise)
 
 signal = X_true + A_true + E_true
+signal = 10 + signal * 40
 
 # Adding missing data
 signal[120:180] = np.nan
 signal[:20] = np.nan
+for i in range(10):
+    signal[i::365] = np.nan
 # signal[80:220] = np.nan
 # mask = np.random.choice(len(signal), round(len(signal) / 20))
 # signal[mask] = np.nan
 
 ```
 
+```python
+import pandas as pd
+df = pd.DataFrame({"signal": signal})
+irn = ImputerRpcaPcp(period=100)
+df_imp = irn.fit_transform(df)
+```
+
+```python
+plt.plot(df_imp["signal"])
+plt.plot(df["signal"])
+
+plt.xlim(0, 200)
+```
+
 ```python tags=[]
 fig = plt.figure(figsize=(15, 8))
 ax = fig.add_subplot(4, 1, 1)

qolmat/imputations/imputers.py

@@ -1698,15 +1698,10 @@ def _transform_element(
         D_scale = (D - means) / stds
         M, A = model.decompose(D_scale, Omega)
         M = M * stds + means
-        A = A * stds + means
 
         M_final = utils.get_shape_original(M, X.shape)
-        A_final = utils.get_shape_original(A, X.shape)
-        X_imputed = M_final + A_final
 
-        df_imputed = pd.DataFrame(
-            X_imputed, index=df.index, columns=df.columns
-        )
+        df_imputed = pd.DataFrame(M_final, index=df.index, columns=df.columns)
         df_imputed = df.where(~df.isna(), df_imputed)
 
         return df_imputed
@@ -1899,7 +1894,6 @@ def _transform_element(
         D_scale = (D - means) / stds
         M, A = model.decompose_on_basis(D_scale, Omega, Q)
         M = M * stds + means
-        A = A * stds + means
 
         M_final = utils.get_shape_original(M, X.shape)