ShapAlgo1

See https://arxiv.org/pdf/1802.03888.pdf - Chapter 3.1, Algorithm 1.

v - vector of node values; = "internal" for internal nodes a,b - left and right node indexes for each internal node t - thresholds for each internal node d - indexes of the features used for splitting in internal nodes r - cover of each node (ie. how many data samples fall in that sub-tree) w - weight, measures the proportion of the training samples matching the conditioning set S fall into each leaf s - set of non-zero indexes in z', ie. known features z' - for each feature, 0 if unknown, 1 if known x - feature values

https://github.com/pkozelka/treeshap