题目
求数据流的中位数
- 设计一个类,包含addNum和findMedian两个方法,findMedian时间复杂度优化到O(1)
方法
- 最简单的方法 - 排序,每次addNum时,元素追加到list末尾;findMedian时快速排序,然后求得中位数;addNum和findMedian的时间复杂度分别是O(n) 和 O(nlogn);提供这个答案明显还不够;如何优化findMedian的时间复杂度呢?
- 维护有序list,二分查找中位数;addNum时保持list有序,二分查找找到小于num的最小元素的索引idx,然后插入新元素到list;addNum和findMedian时间复杂度分别为O(n) O(logn)
- 维护两个heap;求中位数,没必要保持所有元素排序;只需要将元素等分为两个堆heap1,heap2,保证all num in heap1 <= all num in heap2;addNum和findMedian时间复杂度分别为O(logn) O(1)
class MedianFinder:
def __init__(self):
"""
initialize your data structure here.
"""
self.heap1 = []
self.heap2 = []
def addNum(self, num: int) -> None:
l1, l2 = len(self.heap1), len(self.heap2)
if l2 == 0 or self.heap2[0] <= num:
heapq.heappush(self.heap2, num)
else:
heapq.heappush(self.heap1, -num)
l1, l2 = len(self.heap1), len(self.heap2)
if l2 - l1 > 1:
item = heapq.heappop(self.heap2)
heapq.heappush(self.heap1, -item)
elif l1 > l2:
item = heapq.heappop(self.heap1)
heapq.heappush(self.heap2, -item)
# print(self.heap1, self.heap2)
def findMedian(self) -> float:
l1, l2 = len(self.heap1), len(self.heap2)
if l1 + l2 == 0:
return 0
if (l1 + l2) % 2 == 1:
return self.heap2[0]
else:
return (-self.heap1[0] + self.heap2[0]) / 2
# Your MedianFinder object will be instantiated and called as such:
# obj = MedianFinder()
# obj.addNum(num)
# param_2 = obj.findMedian()
- 这个题目前做到了一遍过