Skip to main content

Table 14 LDA Geometric shapes

From: Deriving image features for autonomous classification from time-series recurrence plots

Analyses Hyper-plane side LD1 LD2 LD3 LD4
STANDARD - - Compact - HU1 - Compact - Compact
- Eccentricity - Solidity - HU1 - HU1
+ - HU1 - Compact - Solidity - Solidity
- Homogeneity - Homogeneity
RQA - - Laminarity - Clustering coefficient - Clustering coefficient - Clustering coefficient
- Transitivity - Determinism - Laminarity
- Recurrence rate
+ - Determinism - Laminarity - Recurrence time 1 - Determinism
- Transitivity - Determinism - Recurrence period density
- Recurrence Rate - Transitivity
- Transitivity
STANDARD & RQA - - Homogeneity - Determinism - Solidity - Recurrence rate
- Clustering coefficient - Compact - Recurrence rate - Transitivity
- Laminarity - Homogeneity - Clustering coefficient - Laminarity
- Clustering coefficient - Laminarity - Recurrence time 1
+ - Eccentricity - HU1 - Determinism - Determinism
- Determinism - Solidity - Compact
- Compact - Recurrence rate - HU1
- HU1 - Transitivity - Transitivity
- Recurrence rate - Laminarity
- Transitivity
- Recurrence Time 1
STANDARD & RQA & ESQ - - HU1 - Determinism - Solidity - Recurrence rate
- Homogeneity - Compact - Recurrence rate - Transitivity
- Clustering coefficient - Homogeneity - Clustering coefficient - Laminarity
- Laminarity - Clustering coefficient - Laminarity - Recurrence Time 1
+ - Eccentricity - HU1 - Determinism - Determinism
- Determinism - Solidity - Compact - HU1
- Recurrence rate - Recurrence rate - HU1
- Transitivity - Transitivity - Transitivity
- Recurrence time 1 - Laminarity
  1. Importance of features included in LDA. Features were considered to be important when their loading reached at least 10 % of the maximum loading on the respective side of the hyperplane, set up by the discriminant roots (LD)