Rates From Leave-One-Out Cross-Validation for Selected Nested Alphabets and Orders
|
|
REG alignment segments |
AR alignment segments |
||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Alphabets and orders |
Correct (TP) |
Unclassified |
Wrong (FN) |
Correct (TN) |
Unclassified |
Wrong (FP) |
||||
| 10 symbols | ||||||||||
| order 1 | 0.758242 | 0.000000 | 0.241758 | 0.865385 | 0.000000 | 0.134615 | ||||
| order 2 | 0.805861 | 0.000000 | 0.194139 | 0.850000 | 0.000000 | 0.150000 | ||||
| order 3 | 0.637363 | 0.000000 | 0.362637 | 0.907692 | 0.000000 | 0.092308 | ||||
| order 4 | 0.065934 | 0.919414 | 0.014652 | 0.042308 | 0.957692 | 0.000000 | ||||
| 9 symbols | ||||||||||
| order 1 | 0.743590 | 0.000000 | 0.256410 | 0.873077 | 0.000000 | 0.126923 | ||||
| order 2 | 0.761905 | 0.000000 | 0.238095 | 0.876923 | 0.000000 | 0.123077 | ||||
| order 3 | 0.703297 | 0.000000 | 0.296703 | 0.773077 | 0.000000 | 0.226923 | ||||
| order 4 | 0.069597 | 0.919414 | 0.010989 | 0.088462 | 0.911538 | 0.000000 | ||||
| 8 symbols | ||||||||||
| order 1 | 0.761905 | 0.000000 | 0.238095 | 0.846154 | 0.000000 | 0.153846 | ||||
| order 2 | 0.809524 | 0.000000 | 0.190476 | 0.800000 | 0.000000 | 0.200000 | ||||
| order 3 | 0.706960 | 0.000000 | 0.293040 | 0.773077 | 0.000000 | 0.226923 | ||||
| order 4 | 0.161172 | 0.835165 | 0.003663 | 0.046154 | 0.953846 | 0.000000 | ||||
| 7 symbols | ||||||||||
| order 1 | 0.750916 | 0.000000 | 0.249084 | 0.846154 | 0.000000 | 0.153846 | ||||
| order 2 | 0.791209 | 0.000000 | 0.208791 | 0.834615 | 0.000000 | 0.165385 | ||||
| order 3 | 0.758242 | 0.000000 | 0.241758 | 0.800000 | 0.000000 | 0.200000 | ||||
| order 4 | 0.293040 | 0.695971 | 0.010989 | 0.123077 | 0.869231 | 0.007692 | ||||
| 6 symbols | ||||||||||
| order 1 | 0.747253 | 0.000000 | 0.252747 | 0.846154 | 0.000000 | 0.153846 | ||||
| order 2 | 0.743590 | 0.000000 | 0.256410 | 0.853846 | 0.000000 | 0.146154 | ||||
| order 3 | 0.783883 | 0.000000 | 0.216117 | 0.765385 | 0.000000 | 0.234615 | ||||
| order 4 | 0.498168 | 0.421245 | 0.080586 | 0.384615 | 0.546154 | 0.069231 | ||||
| 5 symbols | ||||||||||
| order 1 | 0.732601 | 0.000000 | 0.267399 | 0.873077 | 0.000000 | 0.126923 | ||||
| order 2 | 0.725275 | 0.000000 | 0.274725 | 0.861538 | 0.000000 | 0.138462 | ||||
| order 3 | 0.681319 | 0.000000 | 0.318681 | 0.838462 | 0.000000 | 0.000000 | ||||
| order 4 | 0.633700 | 0.000000 | 0.366300 | 0.792308 | 0.000000 | 0.207692 | ||||
| order 5
|
0.340659
|
0.351648
|
0.307692
|
0.584615
|
0.361538
|
0.053846
|
||||
-
Thinking of REG as the category to be recognized, the Correct and Wrong columns are also labeled as TP (true positive) FN (false negative) for REG, and TN (true negative) and FP (false positive) for AR. When an order is reached that gives high rates of unclassified elements—over-fitting—larger orders, for which such rates become even higher, are not listed. The 10-symbol alphabet and order 2 (in bold) are the ones used for the 3-way RP and 3-way LRP scores.











