ImputeGAP documentation¶

ImputeGAP is a unified framework for imputation algorithms that provides a narrow-waist interface between algorithm evaluation and parameterization for datasets issued from various domains ranging from neuroscience, medicine, climate to energy.

The interface provides advanced imputation algorithms, construction of various missing values patterns, and different evaluation metrics. In addition, the framework offers support for AutoML parameterization techniques, feature extraction, and, potentially, analysis of feature impact using SHAP. The framework should allow a straightforward integration of new algorithms, datasets, and metrics.

Data Format¶

If you use your own datasets, please make sure your data satisfies the following conditions.

Note

Data Type: Must be a numpy.ndarray.
Structure: The data should be a 2D matrix.
Rows: Each row represents a set of values (e.g., Timestamp #1).
Columns: Each column corresponds to a different series (e.g., Temperature).
Row Separator: Rows should be separated by a carriage return ("\n").
Column Separator: Values within columns should be separated by a space (" ").
Missing Values: Can be detected using numpy.isnan().
Shape: The numpy shape of the matrix imported is (series, values)
Example: Dataset Example - The shape of the time series is (10, 25), the number of series is 10 and the number of values is 25.

Algorithms¶

ALGORITHMS	FAMILIES	CONF
CDRec	Matrix Completion	KAIS’20
IterativeSVD	Matrix Completion	BIOINFORMATICS’01
GROUSE	Matrix Completion	PMLR’16
ROSL	Matrix Completion	CVPR’14
SPIRIT	Matrix Completion	VLDB’05
SoftImpute	Matrix Completion	JMLR’10
SVT	Matrix Completion	SIAM J. OPTIM’10
TRMF	Matrix Completion	NeurIPS’16
ST-MVL	Pattern Search	IJCAI’16
DynaMMo	Pattern Search	KDD’09
TKCM	Pattern Search	EDBT’17
IIM	Machine Learning	ICDE ‘19
XGBI	Machine Learning	KDD’16
Mice	Machine Learning	Statistical Software’11
MissForest	Machine Learning	BioInformatics’11
KNNImpute	Statistics	native
Interpolation	Statistics	native
Min Impute	Statistics	native
Mean Impute	Statistics	native
Mean Impute By Series	Statistics	native
MRNN	Deep Learning	IEEE Trans on BE’19
BRITS	Deep Learning	NeurIPS’18
DeepMVI	Deep Learning	PVLDB’21
MPIN	Deep Learning	PVLDB’24
PriSTI	Deep Learning	ICDE’23
MissNet	Deep Learning	KDD’24
GAIN	Deep Learning	ICML’18
GRIN	Deep Learning	ICLR’22
BayOTIDE	Deep Learning	PMLR’24
HKMF-T	Deep Learning	TKDE’21

API¶

`imputegap.recovery.manager`
`imputegap.recovery.imputation`
`imputegap.recovery.optimization`
`imputegap.recovery.explainer`
`imputegap.recovery.evaluation`
`imputegap.recovery.benchmark`
`imputegap.tools.utils`

ImputeGAP documentation¶

Data Format¶

Algorithms¶

API¶

TREE¶