Let D=(xi,yi)i=1n be a dataset of n input-output pairs lying in Rp×[−M,M] for M∈R. Describe the random-forest algorithm as applied to D using decision trees (T^(b))b=1B to produce a fitted regression function frf. [You need not explain in detail the construction of decision trees, but should describe any modifications specific to the random-forest algorithm.]
Briefly explain why for each x∈Rp and b=1,…,B, we have T^(b)(x)∈[−M,M].
State the bounded-differences inequality.
Treating D as deterministic, show that with probability at least 1−δ,