Features dimension | Interaction order | Memory requirements for interactions | Memory efficiency |
---|
binomialF | Other methods of Table 1 |
---|
10 | 2 | N × 10 | N × 55 | ~ 5 |
3 | N × 175 | ~ 17 |
100 | 2 | N × 100 | N × 5050 | ~ 50 |
3 | N × 166,750 | ~ 1700 |
1000 | 2 | N × 1000 | N × 500,500 | ~ 500 |
3 | N × 166,667,500 | ~ 170,000 |
- The improvement is on the orders of magnitude in 2-way and 3-way interactions when compared to other methods of Table 1. One advantage of the binomialRF algorithm is that it can screen for sets of gene interactions in a memory efficient manner by only requiring a constant-sized matrix whereas the current state of the art requires the predictor matrix to increase in size in a combinatoric fashion to screen for interactions. Memory efficiency is defined by \( \raisebox{1ex}{$\mathrm{Dim}\ \left(\otimes {X}_{i=1}^K\right)$}\!\left/ \!\raisebox{-1ex}{$ Dim(X)$}\right. \), and interaction memory requirements are defined by the number of columns required to map all k-way interactions