Motif Discovery in Functional Data
Add Additive Error to Motif
Add Motif to Base Curve
cluster_candidate_motifs_plot
Cluster Candidate Motifs
Compare Nodes
Functional Motif Discovery
Check Fits
Dissimilarity Index for Multidimensional Curves
Domain Length Calculation for Curves
Find Minimum Dissimilarity
Find Occurrences of a Motif
Generate Coefficients
Custom mapply Function for Parallel Processing
Resample Vector
.select_domain
Transform List Structure
Transform to Matrix
Filter Candidate Motifs
Find Recommended Path in a Tree Structure
funMoDisco: Motif Discovery in Functional Data
Generate Background Curve
Generate Curve Vector from FD Object
Generate Functional Curves with Embedded Motifs
Generate Functional Curves with Embedded Motifs
Get Accolites for a Given Leaf Label
Generate Minimum Dendrogram from Hierarchical Clustering
Get Parent Nodes from a Given Node
Get Complete Paths from a Dendrogram
Initial Checks for ProbKMA
Plot Motif Search Results
Motif Search in Curves
motifSimulationS4Class
motifSimulationApp: A Shiny-Based GUI for Motif Simulation
Create motifSimulation Object
Pad a Matrix to a Specified Number of Rows
Plot Embedded Motifs in Functional Curves
Plot Embedded Motifs in Functional Curves
Plot the Results of probKMA
Filter Motifs from probKMA Results Based on Silhouette and Size Thresh...
Plot Silhouette Index from probKMA Results
Wrapper for the Probabilistic K-means Algorithm (ProbKMA)
ProbKMA Class
Recommend Node from a Numeric Vector
to_motifDiscovery
to_motifDiscovery
Efficiently implementing two complementary methodologies for discovering motifs in functional data: ProbKMA and FunBIalign. Cremona and Chiaromonte (2023) "Probabilistic K-means with Local Alignment for Clustering and Motif Discovery in Functional Data" <doi:10.1080/10618600.2022.2156522> is a probabilistic K-means algorithm that leverages local alignment and fuzzy clustering to identify recurring patterns (candidate functional motifs) across and within curves, allowing different portions of the same curve to belong to different clusters. It includes a family of distances and a normalization to discover various motif types and learns motif lengths in a data-driven manner. It can also be used for local clustering of misaligned data. Di Iorio, Cremona, and Chiaromonte (2023) "funBIalign: A Hierarchical Algorithm for Functional Motif Discovery Based on Mean Squared Residue Scores" <doi:10.48550/arXiv.2306.04254> applies hierarchical agglomerative clustering with a functional generalization of the Mean Squared Residue Score to identify motifs of a specified length in curves. This deterministic method includes a small set of user-tunable parameters. Both algorithms are suitable for single curves or sets of curves. The package also includes a flexible function to simulate functional data with embedded motifs, allowing users to generate benchmark datasets for validating and comparing motif discovery methods.