المدربA window function computes, for each row, a value over a set of related rows — the rows of the same group, ordered, possibly limited to a frame such as the last rows — without collapsing the rows as an aggregation does: a rolling mean per symbol, a rank within a minute, the previous row’s value (a lag).
أمثلة
Example 10.8 (Two sets of bars)
On one symbol-day of synthetic trades (375 trades, stamped to the second as some vendors deliver them), five-minute bars labelled and closed on the left and on the right differ in volume in 4 of the 74 intervals: two trades fall exactly on a five-minute boundary, and each moves its volume, 600 shares in all, from one bar to the next. The two sets also carry labels five minutes apart for the same interval. Filling empty bars adds 4 bars to the 74, and the day’s bar returns become 77 instead of 73: a return computed across a dropped bar spans ten minutes and is compared, in a rolling window, with five-minute returns.