From: G-Tric: generating three-way synthetic datasets with triclustering solutions
Properties | Dataset 1 | Dataset 2 | Dataset 3 | Dataset 4 | Dataset 5 |
---|---|---|---|---|---|
Dataset | |||||
Data type | Real valued | Real valued | Real valued | Binary | Real valued |
Dimensions | 7679 × 13 × 14 | 3200 × 30 × 28 | 20 × 494 × 94 | 51 × 924 × 2844 | 28 × 20 × 365 |
Alphabet | [0, 500] | [5, 1000] | [−5, 5] | 0, 1 | [−10, 30] |
Background | Uniform | Norma l(500, 150) | Uniform | Discrete (0.7, 0.3) | Normal (14, 7) |
Missings | 0% | 20% | 0% | 0% | 15% |
Noise | 0% | 10% | 20% | 0% | 20% |
Errors | 0% | 10% | 15% | 0% | 5% |
Triclusters | |||||
Number | 7 | 10 | 5 | 7000 | 128 |
Dimensions | U(80, 400) × U(2, 4) × U(3, 13) | U(100, 500) × U(10, 20) × U(5, 15) | U(5, 15) × U(50, 200) × U(15, 50) | U(5, 8) × U(20, 70) × U(100, 400) | U(4, 4) × U(4, 4) × U(8, 8) |
Contiguity | No | No | No | No | No |
Patterns | All types | All types | All types | Constant | All types |
Missings | 0% | 10% | 0% | 0% | 5% |
Noise | 0% | 15% | 10% | 0% | 10% |
Errors | 0% | 5% | 5% | 0% | 2% |
Noise deviation | 0 | 2 | 1 | 0 | 1 |
Overlapping | |||||
Plaid coherency | No overlapping | Additive | Additive | None | No overlapping |
% Overlapping trics | 0% | 40% | 100% | 80% | 0% |
Max. interactions | 0 | 2 | 3 | 300 | 0 |
% Overlapping elems. | 0% | 50% | 40% | 70% | 0% |
Restrictions on rows/columns/contexts | 0%/0%/0% | 100%/100%/100% | 100%/100%/100% | 100%/80%/80% | 0%/0%/0% |