Support Page for "Extracting Statistical Graph Features for Accurate and Efficient Time Series Classification"

Page may load slowly, as images are large in size (up to a few megabytes). Please use a desktop browser (Chrome or Safari).

We show examples of how features extracted from MVG can help produce visual cues about classification results. Note that all samples in the test dataset are drawn as line plots in per-class charts. Only ten most important features are shown in scatter matrices.

Dataset: Computers

All samples from test dataset

Scatter matrix of ten most important MVG features


Dataset: FordA

All samples from test dataset

Scatter matrix of ten most important MVG features


Dataset: FordB

All samples from test dataset

Scatter matrix of ten most important MVG features


Dataset: Meat

All samples from test dataset

Scatter matrix of ten most important MVG features


Dataset: SmallKitchenAppliances

All samples from test dataset

Scatter matrix of ten most important MVG features


Dataset: Worms

All samples from test dataset

Scatter matrix of ten most important MVG features


Dataset: WormsTwoClass

All samples from test dataset

Scatter matrix of ten most important MVG features