CDMD : A Controlled Dataset for Time Series Motif Discovery

Short description CDMD dataset is generated for the systematic evaluation of time series motif discovery methods. It is based on series from the UCR archive, but with controlled motif characteristics (length, similarity, heterogeneity).It is designed to enable reproducible and comparative evaluation of motif discovery algorithms, especially in heterogeneous scenarios. Dataset The dataset is organized into 5 groups reflecting varying motif characteristics: Group DescriptionG1 Small radius, small length motifsG2 Large radius, small length motifsG3 Small radius, large length motifsG4 Large radius, large length motifsG5 Mixed heterogeneous motifs Usage notes To use the dataset, first unzip the archive and follow the instructions provided in the README file. Please cite this Zenodo record (using the DOI) in any publication or software that makes use of these data or scripts. The data and associated scripts are packaged in a ZIP archive; please unzip this archive to access the directory structure described in the README file.