Table 8 Rich dataset’s settings
From: G-Tric: generating three-way synthetic datasets with triclustering solutions
Properties | Dataset R | Dataset S | Dataset B | Dataset C |
|---|---|---|---|---|
Dataset | ||||
Data type | Real valued | Symbolic | Binary | Integer |
Dimensions | 1000 × 100 × 100 | 1000 × 100 × 100 | 1000 × 100 × 100 | 1000 × 100 × 100 |
Alphabet | [−100, 100] | {1, 2, 3, 4, 5} | {0, 1} | [0, 100] |
Background | Normal (0, 30) | Discrete (0.1, 0.15, 0.3, 0.3, 0.15) | Uniform | Uniform |
Missings | 2% | 2% | 2% | 2% |
Noise | 10% | 10% | 10% | 10% |
Errors | 5% | 5% | 0% | 5% |
Triclusters | ||||
Number | 30 | 30 | 30 | 30 |
Dimensions | U(30, 50) × U(5, 10) × U(3, 5) | U(30, 50) × U(5, 10) × U(3, 5) | U(30, 50) × U(5, 10) × U(3, 5) | U(30, 50) × U(5, 10) × U(3, 5) |
Contiguity | No | No | No | On contexts |
Patterns | All types | Order preserving, constant | Contant | All types |
Missings | 2% | 2% | 2% | 2% |
Noise | 10% | 10% | 10% | 10% |
Errors | 5% | 5% | 0% | 5% |
Noise deviation | 2 | 1 | 1 | 2 |
Overlapping | ||||
Plaid coherency | Additive | None | None | Multiplicative |
% Overlapping trics | 50% | 40% | 60% | 60% |
Max. interactions | 3 | 2 | 3 | 4 |
% Overlapping elems. | 60% | 50% | 80% | 70% |
Restrictions on rows/columns/contexts | 100%/100%/100% | 100%/100%/100% | 100%/100%/100% | 100%/100%/100% |