Naterq pm rev rel

.github/workflows/pytest_imputation_beta.yml

@@ -35,7 +35,6 @@ jobs:
         python -m pytest ./tests/test_imputation_tkcm.py
         python -m pytest ./tests/test_imputation_deepmvi.py
         python -m pytest ./tests/test_imputation_brits.py
-        python -m pytest ./tests/test_imputation_mpin.py
         python -m pytest ./tests/test_imputation_pristi.py
         python -m pytest ./tests/test_imputation_grin.py
         python -m pytest ./tests/test_imputation_gain.py

README.md

@@ -20,7 +20,7 @@ Access to commonly used datasets in time series research (Datasets).
 <br>
 
 ![Python](https://img.shields.io/badge/Python-v3.12-blue) 
-![Release](https://img.shields.io/badge/Release-v1.0.6-brightgreen) 
+![Release](https://img.shields.io/badge/Release-v1.0.7-brightgreen) 
 ![License](https://img.shields.io/badge/License-GPLv3-blue?style=flat&logo=gnu)
 ![Coverage](https://img.shields.io/badge/Coverage-93%25-brightgreen)
 ![PyPI](https://img.shields.io/pypi/v/imputegap?label=PyPI&color=blue)
@@ -40,13 +40,13 @@ Access to commonly used datasets in time series research (Datasets).
 # List of available imputation algorithms
 | **Family**         | **Algorithm**             | **Venue -- Year**            |
 |--------------------|---------------------------|------------------------------|
-| Deep Learning      | BITGraph [[32]](#ref32)   | ICLR -- 2024                 |
+| Deep Learning      | BitGraph [[32]](#ref32)   | ICLR -- 2024                 |
 | Deep Learning      | BayOTIDE [[30]](#ref30)   | PMLR -- 2024                 |
-| Deep Learning      | MPIN [[25]](#ref25)       | PVLDB -- 2024                |
 | Deep Learning      | MissNet [[27]](#ref27)    | KDD -- 2024                  |
-| Deep Learning      | PriSTI [[26]](#ref26)     | ICDE -- 2023                 |
+| Deep Learning      | MPIN [[25]](#ref25)       | PVLDB -- 2024                |
+| Deep Learning      | PRISTI [[26]](#ref26)     | ICDE -- 2023                 |
 | Deep Learning      | GRIN [[29]](#ref29)       | ICLR -- 2022                 |
-| Deep Learning      | HKMF-T [[31]](#ref31)     | TKDE -- 2021                 |
+| Deep Learning      | HKMF_T [[31]](#ref31)     | TKDE -- 2021                 |
 | Deep Learning      | DeepMVI [[24]](#ref24)    | PVLDB -- 2021                |
 | Deep Learning      | MRNN [[22]](#ref22)       | IEEE Trans on BE -- 2019     |
 | Deep Learning      | BRITS [[23]](#ref23)      | NeurIPS -- 2018              |
@@ -60,18 +60,18 @@ Access to commonly used datasets in time series research (Datasets).
 | Matrix Completion  | SPIRIT [[5]](#ref5)       | VLDB -- 2005                 |
 | Matrix Completion  | IterativeSVD [[2]](#ref2) | BIOINFORMATICS -- 2001       |
 | Pattern Search     | TKCM [[11]](#ref11)       | EDBT -- 2017                 |
-| Pattern Search     | ST-MVL [[9]](#ref9)       | IJCAI -- 2016                |
+| Pattern Search     | STMVL [[9]](#ref9)        | IJCAI -- 2016                |
 | Pattern Search     | DynaMMo [[10]](#ref10)    | KDD -- 2009                  |
 | Machine Learning   | IIM [[12]](#ref12)        | ICDE -- 2019                 |
-| Machine Learning   | XGBI [[13]](#ref13)       | KDD -- 2016                  |
-| Machine Learning   | Mice [[14]](#ref14)       | Statistical Software -- 2011 |
+| Machine Learning   | XGBOOST [[13]](#ref13)    | KDD -- 2016                  |
+| Machine Learning   | MICE [[14]](#ref14)       | Statistical Software -- 2011 |
 | Machine Learning   | MissForest [[15]](#ref15) | BioInformatics -- 2011       |
 | Statistics         | KNNImpute                 | -                            |
 | Statistics         | Interpolation             | -                            |
-| Statistics         | Min Impute                | -                            |
-| Statistics         | Zero Impute               | -                            |
-| Statistics         | Mean Impute               | -                            |
-| Statistics         | Mean Impute By Series     | -                            |
+| Statistics         | MinImpute                 | -                            |
+| Statistics         | ZeroImpute                | -                            |
+| Statistics         | MeanImpute                | -                            |
+| Statistics         | MeanImputeBySeries        | -                            |
 
 ---
 
@@ -155,7 +155,7 @@ ts.normalize(normalizer="z_score")
 
 # plot and print a subset of time series
 ts.plot(input_data=ts.data, nbr_series=9, nbr_val=100, save_path="./imputegap_assets")
-ts.print(nbr_series=9, nbr_val=100)
+ts.print(nbr_series=9, nbr_val=20)
 ```
 
 ---
@@ -235,8 +235,7 @@ imputer.score(ts.data, imputer.recov_data)
 ts.print_results(imputer.metrics)
 
 # plot the recovered time series
-ts.plot(input_data=ts.data, incomp_data=ts_m, recov_data=imputer.recov_data, nbr_series=9, subplot=True,
-        save_path="./imputegap_assets")
+ts.plot(input_data=ts.data, incomp_data=ts_m, recov_data=imputer.recov_data, nbr_series=9, subplot=True, algorithm=imputer.algorithm, save_path="./imputegap_assets")
 ```
 
 ---
@@ -270,9 +269,16 @@ imputer = Imputation.MatrixCompletion.CDRec(ts_m)
 # use Ray Tune to fine tune the imputation algorithm
 imputer.impute(user_def=False, params={"input_data": ts.data, "optimizer": "ray_tune"})
 
-# compute and print the imputation metrics
+# compute the imputation metrics with optimized parameter values
 imputer.score(ts.data, imputer.recov_data)
-ts.print_results(imputer.metrics)
+
+# compute the imputation metrics with default parameter values
+imputer_def = Imputation.MatrixCompletion.CDRec(ts_m).impute()
+imputer_def.score(ts.data, imputer_def.recov_data)
+
+# print the imputation metrics with default and optimized parameter values
+ts.print_results(imputer_def.metrics, text="Imputation metrics with default parameter values")
+ts.print_results(imputer.metrics, text="Imputation metrics with optimized parameter values")
 
 # plot the recovered time series
 ts.plot(input_data=ts.data, incomp_data=ts_m, recov_data=imputer.recov_data, nbr_series=9, subplot=True,
@@ -357,7 +363,7 @@ imputer = Imputation.MatrixCompletion.CDRec(ts_m)
 imputer.impute()
 
 # compute and print the downstream results
-downstream_config = {"task": "forecast", "model": "hw-add"}
+downstream_config = {"task": "forecast", "model": "hw-add", "comparator": "ZeroImpute"}
 imputer.score(ts.data, imputer.recov_data, downstream=downstream_config)
 ts.print_results(imputer.downstream_metrics, algorithm=imputer.algorithm)
 ```
@@ -369,7 +375,7 @@ ts.print_results(imputer.downstream_metrics, algorithm=imputer.algorithm)
 
 ## Benchmark
 
-ImputeGAP can serve as a common test-bed for comparing the effectiveness and efficiency of time series imputation algorithms[[33]](#ref33) . Users have full control over the benchmark by customizing various parameters, including the list of datasets to evaluate, the algorithms to compare, the choice of optimizer to fine-tune the algorithms on the chosen datasets, the missingness patterns, and the range of missing rates.
+ImputeGAP can serve as a common test-bed for comparing the effectiveness and efficiency of time series imputation algorithms[[33]](#ref33) . Users have full control over the benchmark by customizing various parameters, including the list of datasets to evaluate, the algorithms to compare, the choice of optimizer to fine-tune the algorithms on the chosen datasets, the missingness patterns, and the range of missing rates. The default metrics evaluated include "RMSE", "MAE", "MI", "Pearson", and the runtime.
 The documentation for the benchmark is described [here](https://imputegap.readthedocs.io/en/latest/benchmark.html).
 
 
@@ -381,22 +387,27 @@ The benchmarking module can be utilized as follows:
 ```python
 from imputegap.recovery.benchmark import Benchmark
 
-save_dir = "./analysis"
-nbr_run = 2
+save_dir = "./imputegap_assets/benchmark"
+nbr_runs = 1
 
-datasets = ["eeg-alcohol", "eeg-reading"]
+datasets = ["eeg-alcohol"]
 
-optimizer = {"optimizer": "ray_tune", "options": {"n_calls": 1, "max_concurrent_trials": 1}}
-optimizers = [optimizer]
+optimizers = ["default_params"]
 
-algorithms = ["MeanImpute", "CDRec", "STMVL", "IIM", "MRNN"]
+algorithms = ["SoftImpute", "KNNImpute"]
 
 patterns = ["mcar"]
 
 range = [0.05, 0.1, 0.2, 0.4, 0.6, 0.8]
 
 # launch the evaluation
-list_results, sum_scores = Benchmark().eval(algorithms=algorithms, datasets=datasets, patterns=patterns, x_axis=range, optimizers=optimizers, save_dir=save_dir, runs=nbr_run)
+list_results, sum_scores = Benchmark().eval(algorithms=algorithms, datasets=datasets, patterns=patterns, x_axis=range, optimizers=optimizers, save_dir=save_dir, runs=nbr_runs)
+```
+
+You can change the optimizer using the following command:
+```python
+optimizer = {"optimizer": "ray_tune", "options": {"n_calls": 1, "max_concurrent_trials": 1}}
+optimizers = [optimizer]
 ```
 
 ---
@@ -407,20 +418,6 @@ To add your own imputation algorithm in Python or C++, please refer to the detai
 
 ---
 
-
-## Articles
-
-
-Mourad Khayati, Alberto Lerner, Zakhar Tymchenko, Philippe Cudre-Mauroux: Mind the Gap: An Experimental Evaluation of Imputation of Missing Values Techniques in Time Series. Proc. VLDB Endow. 13(5): 768-782 (2020)
-
-Mourad Khayati, Quentin Nater, Jacques Pasquier: ImputeVIS: An Interactive Evaluator to Benchmark Imputation Techniques for Time Series Data. Proc. VLDB Endow. 17(12): 4329-4332 (2024)
-
----
-
-
-
-
-
 ## Citing
 
 If you use ImputeGAP in your research, please cite the paper:

build/lib/imputegap/__init__.py

@@ -1 +1 @@
-__version__ = "1.0.5"
+__version__ = "1.0.7"

build/lib/imputegap/algorithms/bayotide.py

@@ -66,6 +66,6 @@ def bay_otide(incomp_data, K_trend=20, K_season=2, n_season=5, K_bias=1, time_sc
 
     end_time = time.time()
     if logs:
-        print(f"\n\t\t> logs, imputation bay_otide - Execution Time: {(end_time - start_time):.4f} seconds\n")
+        print(f"\n\t> logs, imputation bay_otide - Execution Time: {(end_time - start_time):.4f} seconds\n")
 
     return recov_data

build/lib/imputegap/algorithms/bit_graph.py

@@ -64,6 +64,6 @@ def bit_graph(incomp_data, node_number=-1, kernel_set=[1], dropout=0.1, subgraph
 
     end_time = time.time()
     if logs:
-        print(f"\n\t\t> logs, imputation bit graph - Execution Time: {(end_time - start_time):.4f} seconds\n")
+        print(f"\n\t> logs, imputation bit graph - Execution Time: {(end_time - start_time):.4f} seconds\n")
 
     return recov_data

build/lib/imputegap/algorithms/brits.py

@@ -49,6 +49,6 @@ def brits(incomp_data, model="brits", epoch=10, batch_size=7, nbr_features=1, hi
 
     end_time = time.time()
     if logs:
-        print(f"\n\t\t> logs, imputation brits - Execution Time: {(end_time - start_time):.4f} seconds\n")
+        print(f"\n\t> logs, imputation brits - Execution Time: {(end_time - start_time):.4f} seconds\n")
 
     return recov_data

build/lib/imputegap/algorithms/cdrec.py

@@ -29,7 +29,7 @@ def native_cdrec(__py_matrix, __py_rank, __py_epsilon, __py_iterations):
     Khayati, M., Cudré-Mauroux, P. & Böhlen, M.H. Scalable recovery of missing blocks in time series with high and low cross-correlations. Knowl Inf Syst 62, 2257–2280 (2020). https://doi.org/10.1007/s10115-019-01421-7
     """
 
-    shared_lib = utils.load_share_lib("lib_cdrec.so")
+    shared_lib = utils.load_share_lib("lib_cdrec")
 
     __py_n = len(__py_matrix);
     __py_m = len(__py_matrix[0]);
@@ -87,7 +87,7 @@ def cdrec(incomp_data, truncation_rank, iterations, epsilon, logs=True, lib_path
 
     """
 
-    print(f"\t\t\t\t(PYTHON) CDRec: ({incomp_data.shape[0]},{incomp_data.shape[1]}) for rank {truncation_rank}, "
+    print(f"(PYTHON) CDRec: ({incomp_data.shape[0]},{incomp_data.shape[1]}) for rank {truncation_rank}, "
           f"epsilon {epsilon}, and iterations {iterations}...")
 
     start_time = time.time()  # Record start time
@@ -98,6 +98,6 @@ def cdrec(incomp_data, truncation_rank, iterations, epsilon, logs=True, lib_path
     end_time = time.time()
 
     if logs:
-        print(f"\n\t\t> logs, imputation cdrec - Execution Time: {(end_time - start_time):.4f} seconds\n")
+        print(f"\n\t> logs, imputation cdrec - Execution Time: {(end_time - start_time):.4f} seconds\n")
 
     return recov_data

build/lib/imputegap/algorithms/cpp_integration.py

@@ -25,7 +25,7 @@ def native_algo(__py_matrix, __py_param):
 
     """
 
-    shared_lib = utils.load_share_lib("to_adapt.so")
+    shared_lib = utils.load_share_lib("to_adapt")
 
     __py_n = len(__py_matrix);
     __py_m = len(__py_matrix[0]);
@@ -79,6 +79,6 @@ def your_algo(contamination, param, logs=True):
     end_time = time.time()
 
     if logs:
-        print(f"\n\t\t> logs, imputation algo - Execution Time: {(end_time - start_time):.4f} seconds\n")
+        print(f"\n\t> logs, imputation algo - Execution Time: {(end_time - start_time):.4f} seconds\n")
 
     return recov_data

build/lib/imputegap/algorithms/deep_mvi.py

@@ -41,6 +41,6 @@ def deep_mvi(incomp_data, max_epoch=1000, patience=2, lr=0.001, logs=True):
 
     end_time = time.time()
     if logs:
-        print(f"\n\t\t> logs, imputation deep mvi - Execution Time: {(end_time - start_time):.4f} seconds\n")
+        print(f"\n\t> logs, imputation deep mvi - Execution Time: {(end_time - start_time):.4f} seconds\n")
 
     return recov_data

build/lib/imputegap/algorithms/dynammo.py

@@ -29,7 +29,7 @@ def native_dynammo(__py_matrix, __py_h, __py_maxIter, __py_fast):
     L. Li, J. McCann, N. S. Pollard, and C. Faloutsos. Dynammo: mining and summarization of coevolving sequences with missing values. In Proceedings of the 15th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, Paris, France, June 28 - July 1, 2009, pages 507–516, 2009.
     """
 
-    shared_lib = utils.load_share_lib("lib_dynammo.so")
+    shared_lib = utils.load_share_lib("lib_dynammo")
 
     __py_n = len(__py_matrix);
     __py_m = len(__py_matrix[0]);
@@ -92,6 +92,6 @@ def dynammo(incomp_data, h, max_iteration, approximation, logs=True, lib_path=No
     end_time = time.time()
 
     if logs:
-        print(f"\n\t\t> logs, imputation DynaMMo - Execution Time: {(end_time - start_time):.4f} seconds\n")
+        print(f"\n\t> logs, imputation DynaMMo - Execution Time: {(end_time - start_time):.4f} seconds\n")
 
     return recov_data

build/lib/imputegap/algorithms/gain.py

@@ -44,6 +44,6 @@ def gain(incomp_data, batch_size=32, hint_rate=0.9, alpha=10, epoch=100, logs=Tr
 
     end_time = time.time()
     if logs:
-        print(f"\n\t\t> logs, imputation gain - Execution Time: {(end_time - start_time):.4f} seconds\n")
+        print(f"\n\t> logs, imputation gain - Execution Time: {(end_time - start_time):.4f} seconds\n")
 
     return recov_data

build/lib/imputegap/algorithms/grin.py

@@ -60,6 +60,6 @@ def grin(incomp_data, d_hidden=32, lr=0.001, batch_size=32, window=10, alpha=10.
 
     end_time = time.time()
     if logs:
-        print(f"\n\t\t> logs, imputation grin - Execution Time: {(end_time - start_time):.4f} seconds\n")
+        print(f"\n\t> logs, imputation grin - Execution Time: {(end_time - start_time):.4f} seconds\n")
 
     return recov_data

build/lib/imputegap/algorithms/grouse.py

@@ -24,7 +24,7 @@ def native_grouse(__py_matrix, __py_rank):
     D. Zhang and L. Balzano. Global convergence of a grassmannian gradient descent algorithm for subspace estimation. In Proceedings of the 19th International Conference on Artificial Intelligence and Statistics, AISTATS 2016, Cadiz, Spain, May 9-11, 2016, pages 1460–1468, 2016.
     """
 
-    shared_lib = utils.load_share_lib("lib_grouse.so")
+    shared_lib = utils.load_share_lib("lib_grouse")
 
     __py_n = len(__py_matrix);
     __py_m = len(__py_matrix[0]);
@@ -80,6 +80,6 @@ def grouse(incomp_data, max_rank, logs=True, lib_path=None):
     end_time = time.time()
 
     if logs:
-        print(f"\n\t\t> logs, imputation GROUSE - Execution Time: {(end_time - start_time):.4f} seconds\n")
+        print(f"\n\t> logs, imputation GROUSE - Execution Time: {(end_time - start_time):.4f} seconds\n")
 
     return recov_data

build/lib/imputegap/algorithms/hkmf_t.py

@@ -48,6 +48,6 @@ def hkmf_t(incomp_data, tags=None, data_names=None, epoch=10, logs=True):
 
     end_time = time.time()
     if logs:
-        print(f"\n\t\t> logs, imputation hkmf_t - Execution Time: {(end_time - start_time):.4f} seconds\n")
+        print(f"\n\t> logs, imputation hkmf_t - Execution Time: {(end_time - start_time):.4f} seconds\n")
 
     return recov_data

build/lib/imputegap/algorithms/iim.py

@@ -45,6 +45,6 @@ def iim(incomp_data, number_neighbor, algo_code, logs=True):
 
     end_time = time.time()
     if logs:
-        print(f"\n\t\t> logs, imputation iim - Execution Time: {(end_time - start_time):.4f} seconds\n")
+        print(f"\n\t> logs, imputation iim - Execution Time: {(end_time - start_time):.4f} seconds\n")
 
     return recov_data

build/lib/imputegap/algorithms/interpolation.py

@@ -30,7 +30,7 @@ def interpolation(incomp_data, method="linear", poly_order=2, logs=True):
 
     """
 
-    print(f"\t\t\t\t(PYTHON) interpolation : ({incomp_data.shape[0]},{incomp_data.shape[1]}) for method {method}"
+    print(f"(PYTHON) interpolation : ({incomp_data.shape[0]},{incomp_data.shape[1]}) for method {method}"
           f", and polynomial order {poly_order}...")
 
     start_time = time.time()  # Record start time
@@ -70,6 +70,6 @@ def interpolation(incomp_data, method="linear", poly_order=2, logs=True):
 
     end_time = time.time()
     if logs:
-        print(f"\n\t\t> logs, imputation with interpolation - Execution Time: {(end_time - start_time):.4f} seconds\n")
+        print(f"\n\t> logs, imputation with interpolation - Execution Time: {(end_time - start_time):.4f} seconds\n")
 
     return recov_data

build/lib/imputegap/algorithms/iterative_svd.py

@@ -25,7 +25,7 @@ def native_iterative_svd(__py_matrix, __py_rank):
     Olga Troyanskaya, Michael Cantor, Gavin Sherlock, Pat Brown, Trevor Hastie, Robert Tibshirani, David Botstein, Russ B. Altman, Missing value estimation methods for DNA microarrays , Bioinformatics, Volume 17, Issue 6, June 2001, Pages 520–525, https://doi.org/10.1093/bioinformatics/17.6.520
     """
 
-    shared_lib = utils.load_share_lib("lib_iterative_svd.so")
+    shared_lib = utils.load_share_lib("lib_iterative_svd")
 
     __py_n = len(__py_matrix);
     __py_m = len(__py_matrix[0]);
@@ -82,6 +82,6 @@ def iterative_svd(incomp_data, truncation_rank, logs=True, lib_path=None):
     end_time = time.time()
 
     if logs:
-        print(f"\n\t\t> logs, imputation iterative svd - Execution Time: {(end_time - start_time):.4f} seconds\n")
+        print(f"\n\t> logs, imputation iterative svd - Execution Time: {(end_time - start_time):.4f} seconds\n")
 
     return recov_data

build/lib/imputegap/algorithms/knn.py

@@ -29,7 +29,7 @@ def knn(incomp_data, k=5, weights="uniform", logs=True):
 
     """
 
-    print(f"\t\t\t\t(PYTHON) KNN: ({incomp_data.shape[0]},{incomp_data.shape[1]}) for k {k}, "
+    print(f"(PYTHON) KNNImpute: ({incomp_data.shape[0]},{incomp_data.shape[1]}) for k {k}, "
           f", and weights {weights}...")
 
     start_time = time.time()  # Record start time
@@ -74,6 +74,6 @@ def knn(incomp_data, k=5, weights="uniform", logs=True):
 
     end_time = time.time()
     if logs:
-        print(f"\n\t\t> logs, imputation knn - Execution Time: {(end_time - start_time):.4f} seconds\n")
+        print(f"\n\t> logs, imputation knn_impute - Execution Time: {(end_time - start_time):.4f} seconds\n")
 
     return recov_data