Given a stream of data points (*), find the p50 and p95.
[** **** * *** ****** **********]p50 = ****
p95 = ******
This problem asks you to compute streaming percentiles such as P50 and P95 from incoming data points. A strong solution should discuss online data structures or algorithms, such as two heaps, balanced search trees, or approximate quantile methods, instead of storing and sorting the entire stream. The key interview focus is how to maintain accurate percentile estimates efficiently under memory and latency constraints.
END
