2. correction of optimization and creation of batches

imputegap/explainer/explainer.py

@@ -8,8 +8,8 @@
 from matplotlib import pyplot as plt
 from sklearn.ensemble import RandomForestRegressor
 
-from imputegap.contamination.contamination import Contamination
-from imputegap.imputation.imputation import Imputation
+from imputegap.recovery.contamination import Contamination
+from imputegap.recovery.imputation import Imputation
 
 
 class Explainer:

imputegap/contamination/README.md → imputegap/recovery/README.md

@@ -1,6 +1,7 @@
 ![My Logo](../../assets/logo_imputegab.png)
 
-# Scenarios
+# CONTAMINATION
+## Scenarios
 <table>
     <tr>
         <td>M</td><td>Number of time series</td>
@@ -23,7 +24,7 @@
     <tr>
         <td>B</td><td>Block size</td>
     </tr>
-</table>
+</table><br />
 
 ### MCAR
 MCAR selects random series and remove block at random positions until a total of W of all points of time series are missing.
@@ -32,7 +33,7 @@ This scenario uses random number generator with fixed seed and will produce the
 <table>
     <tbody>Definition</tbody>
     <tr>
-        <td>N</td><td>10 - 100%</td>
+        <td>N</td><td>MAX</td>
     </tr>
     <tr>
         <td>M</td><td>MAX</td>
@@ -49,4 +50,31 @@ This scenario uses random number generator with fixed seed and will produce the
     <tr>
         <td>B</td><td>2 - 20</td>
     </tr>
- </table>
+ </table>
+
+<br />
+
+### MISSING PERCENTAGE
+MISSING PERCENTAGE selects of percent of series to contaminate from the first to the last with a desired percentage of missing value to remove.
+
+<table>
+    <tbody>Definition</tbody>
+    <tr>
+        <td>N</td><td>MAX</td>
+    </tr>
+    <tr>
+        <td>M</td><td>MAX</td>
+    </tr>
+    <tr>
+        <td>R</td><td>1 - 100%</td>
+    </tr>
+    <tr>
+        <td>S</td><td>1 - 100%</td>
+    </tr>
+    <tr>
+        <td>W</td><td>(N-P) * R</td>
+    </tr>
+    <tr>
+        <td>B</td><td>R</td>
+    </tr>
+ </table><br />

imputegap/contamination/contamination.py → imputegap/recovery/contamination.py

@@ -1,9 +1,27 @@
-import math
 import numpy as np
 
 
 class Contamination:
 
+    def verification_limitation(percentage, low_limit=0.01, high_limit=1.0):
+        """
+        Format the percentage given by the user.
+        :param percentage: The percentage to be checked.
+        :param low_limit: The lower limit of the acceptable percentage range.
+        :param high_limit: The upper limit of the acceptable percentage range.
+        :return: Adjusted percentage.
+        """
+        if low_limit <= percentage <= high_limit:
+            return percentage  # No modification needed
+
+        elif 1 <= percentage <= 100:
+            print(f"The percentage {percentage} is between 1 and 100. Dividing by 100 to convert to a decimal.")
+            return percentage / 100
+
+        else:
+            print("The percentage", percentage, "is out of the acceptable range", low_limit, "-", high_limit, ".")
+            return percentage
+
     def format_selection(ts, selection):
         """
         Format the selection of series based on keywords
@@ -59,6 +77,10 @@ def scenario_mcar(ts, series_impacted=0.2, missing_rate=0.2, block_size=10, prot
         if use_seed:
             np.random.seed(seed)
 
+        missing_rate = Contamination.verification_limitation(missing_rate)
+        series_impacted = Contamination.verification_limitation(series_impacted)
+        protection = Contamination.verification_limitation(protection)
+
         ts_contaminated = ts.copy()
         M, _ = ts_contaminated.shape
 
@@ -128,6 +150,10 @@ def scenario_missing_percentage(ts, series_impacted=0.2, missing_rate=0.2, prote
         ts_contaminated = ts.copy()
         M, _ = ts_contaminated.shape
 
+        missing_rate = Contamination.verification_limitation(missing_rate)
+        series_impacted = Contamination.verification_limitation(series_impacted)
+        protection = Contamination.verification_limitation(protection)
+
         nbr_series_impacted = int(np.ceil(M * series_impacted))
 
         print("\n\nMISSING PERCENTAGE contamination has been called with :"

imputegap/imputation/imputation.py → imputegap/recovery/imputation.py

@@ -1,14 +1,12 @@
 import re
-import os
-import toml
 from imputegap.algorithms.cdrec import cdrec
 from imputegap.algorithms.iim import iim
 from imputegap.algorithms.min_impute import min_impute
 from imputegap.algorithms.mrnn import mrnn
 from imputegap.algorithms.stmvl import stmvl
 from imputegap.algorithms.zero_impute import zero_impute
-from imputegap.evaluation.evaluation import Evaluation
-from imputegap.manager import utils
+from imputegap.tools.evaluation import Evaluation
+from imputegap.tools import utils
 
 
 class Imputation:

imputegap/manager/manager.py → imputegap/recovery/manager.py

@@ -145,11 +145,12 @@ def normalization_z_score(self, ts):
 
         return z_scores
 
-    def plot(self, ts_type="ground_truth", title='Time Series Data', save_path="", limitation=10, size=(16, 8), display=True, colors=['dimgrey', 'plum', 'lightblue', 'mediumseagreen', 'khaki']):
+    def plot(self, ts_type="gt", title='Time Series Data', save_path="", limitation=10, size=(16, 8), display=True, colors=['dimgrey', 'plum', 'lightblue', 'mediumseagreen', 'khaki']):
         """
         Plot a chosen time series
         @author Quentin Nater
 
+        :param ts_type : ("gt", "contamination", "imputation"), type of time series to print | default "gt"
         :param title: title of the plot
         :param save_path : path to save locally the plot
         :param limitation: number of series displayed inside the plot
@@ -165,29 +166,24 @@ def plot(self, ts_type="ground_truth", title='Time Series Data', save_path="", l
         if limitation == 0:
             limitation = self.ts.shape[0]
 
-        if ts_type == "ground_truth":
+        if ts_type == "gt":
             for i in range(self.ts.shape[0]):
                 plt.plot(np.arange(self.ts.shape[1]), self.ts[i, :], label=f'Series {i + 1}')
                 number_of_series += 1
 
                 if number_of_series == limitation:
                     break
 
-        elif ts_type == "ground_truth_normalized":
-            for i in range(self.normalized_ts.shape[0]):
-                plt.plot(np.arange(self.normalized_ts.shape[1]), self.normalized_ts[i, :], label=f'Series {i + 1}')
-                number_of_series += 1
-
-                if number_of_series == limitation:
-                    break
 
         elif ts_type == "contamination":
             for i in range(self.ts.shape[0]):
                 color = colors[i % len(colors)]
 
-                plt.plot(np.arange(self.ts.shape[1]), self.ts[i, :], 'r--', label=f'Series {i + 1}-GT')
+                if np.isnan(self.ts_contaminate[i, :]).any():
+                    plt.plot(np.arange(self.ts.shape[1]), self.ts[i, :], 'r--', label=f'Series {i + 1}')
+
                 plt.plot(np.arange(self.ts_contaminate.shape[1]), self.ts_contaminate[i, :], linewidth=2.5,
-                         color=color, linestyle='-', label=f'Series {i + 1}-MV')
+                         color=color, linestyle='-', label=f'Series {i + 1}-CNT')
 
                 number_of_series += 1
                 if number_of_series == limitation:

imputegap/optimization/bayesian_optimization.py → imputegap/recovery/optimization.py

@@ -1,13 +1,14 @@
 import os
-
+import toml
 import numpy as np
+
+
 import skopt
-import toml
 from skopt.space import Integer
 from skopt.utils import use_named_args
 
-from imputegap.imputation.imputation import Imputation
-from imputegap.optimization.algorithm_parameters import SEARCH_SPACES
+from imputegap.recovery.imputation import Imputation
+from imputegap.tools.algorithm_parameters import SEARCH_SPACES
 
 # Define the search space for each algorithm separately
 search_spaces = SEARCH_SPACES

imputegap/runner_contamination.py

@@ -0,0 +1,32 @@
+from imputegap.recovery.contamination import Contamination
+from imputegap.recovery.imputation import Imputation
+from imputegap.recovery.manager import TimeSeries
+from imputegap.recovery.optimization import Optimization
+from imputegap.tools.utils import display_title
+from imputegap.tools import utils
+import os
+
+
+def check_block_size(filename):
+    if "test" in filename:
+        return (2, 2)
+    else:
+        return (10, 10)
+
+
+if __name__ == '__main__':
+
+    display_title()
+
+    filename = "eeg"
+
+    file_path = os.path.join("./dataset/", filename + ".txt")
+    gap = TimeSeries(data=file_path, normalization="z_score", limitation_values=100)
+
+    block_size, plot_limit = check_block_size(filename)
+
+    gap.ts_contaminate = Contamination.scenario_missing_percentage(ts=gap.ts, series_impacted=0.1, missing_rate=0.4, protection=0.1, use_seed=True, seed=42)
+    gap.print(limitation=10)
+    gap.plot(ts_type="contamination", title="test", save_path="assets", limitation=plot_limit, display=True)
+
+    print("\n", "_"*95, "end")

imputegap/runner_explainer.py

@@ -1,16 +1,8 @@
-from imputegap.contamination.contamination import Contamination
-from imputegap.imputation.imputation import Imputation
-from imputegap.manager.manager import TimeSeries
+from imputegap.recovery.manager import TimeSeries
 from imputegap.explainer.explainer import Explainer
-import os
-
+from imputegap.tools.utils import display_title
 
-def display_title(title="Master Thesis", aut="Quentin Nater", lib="ImputeGAP", university="University Fribourg - exascale infolab"):
-    print("=" * 100)
-    print(f"{title} : {aut}")
-    print("=" * 100)
-    print(f"    {lib} - {university}")
-    print("=" * 100)
+import os
 
 
 def check_block_size(filename):

imputegap/runner_imputation.py

@@ -1,19 +1,11 @@
-from imputegap.contamination.contamination import Contamination
-from imputegap.imputation.imputation import Imputation
-from imputegap.manager import utils
-from imputegap.manager.manager import TimeSeries
+from imputegap.recovery.contamination import Contamination
+from imputegap.recovery.imputation import Imputation
+from imputegap.recovery.manager import TimeSeries
+from imputegap.recovery.optimization import Optimization
+from imputegap.tools.utils import display_title
+from imputegap.tools import utils
 import os
 
-from imputegap.optimization.bayesian_optimization import Optimization
-
-
-def display_title(title="Master Thesis", aut="Quentin Nater", lib="ImputeGAP", university="University Fribourg - exascale infolab"):
-    print("=" * 100)
-    print(f"{title} : {aut}")
-    print("=" * 100)
-    print(f"    {lib} - {university}")
-    print("=" * 100)
-
 
 def check_block_size(filename):
     if "test" in filename:
@@ -36,7 +28,7 @@ def check_block_size(filename):
     block_size, plot_limit = check_block_size(filename)
 
     gap.print(limitation=5)
-    gap.plot(title="test", save_path="assets", limitation=0, display=False)
+    gap.plot(ts_type="gt", title="test", save_path="assets", limitation=0, display=False)
 
     gap.ts_contaminate = Contamination.scenario_mcar(ts=gap.ts, series_impacted=0.4, missing_rate=0.4, block_size=block_size, protection=0.1, use_seed=True, seed=42)
     gap.print(limitation=10)

imputegap/runner_optimization.py

@@ -1,16 +1,9 @@
-from imputegap.contamination.contamination import Contamination
-from imputegap.manager.manager import TimeSeries
-import os
-
-from imputegap.optimization.bayesian_optimization import Optimization
+from imputegap.recovery.contamination import Contamination
+from imputegap.recovery.manager import TimeSeries
+from imputegap.recovery.optimization import Optimization
+from imputegap.tools.utils import display_title
 
-
-def display_title(title="Master Thesis", aut="Quentin Nater", lib="ImputeGAP", university="University Fribourg - exascale infolab"):
-    print("=" * 100)
-    print(f"{title} : {aut}")
-    print("=" * 100)
-    print(f"    {lib} - {university}")
-    print("=" * 100)
+import os
 
 
 def check_block_size(filename):

imputegap/optimization/algorithm_parameters.py → imputegap/tools/algorithm_parameters.py

@@ -8,8 +8,6 @@
 
 # IIM parameters
 IIM_LEARNING_NEIGHBOR_RANGE = [i for i in range(1, 100)]  # Test up to 100 learning neighbors
-# IIM_ADAPTIVE_RANGE = [True, False]  # Test with and without adaptive learning
-# IIM_METRIC_RANGE = ['euclidean', 'cosine']  # Test with euclidean and cosine distance, could be more
 
 # MRNN parameters
 MRNN_LEARNING_RATE_CHANGE = np.logspace(-6, 0, num=20)  # log scale for learning rate
@@ -45,11 +43,4 @@
     'iim': ['learning_neighbors'],
     'mrnn': ['hidden_dim', 'learning_rate', 'iterations', 'seq_len' ],
     'stmvl': ['window_size', 'gamma', 'alpha']
-}
-
-DEFAULT_PARAMS = {
-    'cdrec': [1, 1e-6, 100],
-    'iim': [10],
-    'mrnn': [10, 0.01, 1000, 7],
-    'stmvl': [2, 0.85, 7]
 }

imputegap/manager/utils.py → imputegap/tools/utils.py

@@ -1,8 +1,15 @@
 import os
-
 import toml
 
 
+def display_title(title="Master Thesis", aut="Quentin Nater", lib="ImputeGAP", university="University Fribourg - exascale infolab"):
+    print("=" * 100)
+    print(f"{title} : {aut}")
+    print("=" * 100)
+    print(f"    {lib} - {university}")
+    print("=" * 100)
+
+
 def get_file_path_dataset(set_name="test"):
     """
     Find the accurate path for loading files of tests

tests/test_contamination_mcar.py

@@ -2,9 +2,9 @@
 import unittest
 import numpy as np
 
-from imputegap.contamination.contamination import Contamination
-from imputegap.manager import utils
-from imputegap.manager.manager import TimeSeries
+from imputegap.recovery.contamination import Contamination
+from imputegap.tools import utils
+from imputegap.recovery.manager import TimeSeries
 
 
 class TestContamination(unittest.TestCase):