题目:295. 数据流的中位数
链接:https://leetcode-cn.com/problems/find-median-from-data-stream/
题目描述:
看到这个题目第一反应,排序就行了嘛!下面直接用排序写:
public class MedianFinder {
int[] arr;
int length;
public MedianFinder() {
arr = new int[100000];
length = 0;
}
public void addNum(int num) {
arr[length ++] = num;
}
public double findMedian() {
Arrays.sort(arr, 0, length);
// int[] arrs = Arrays.copyOf(arr, length);
// Arrays.sort(arrs);
if (length == 0) return 0;
if (length % 2 == 0) {
return (arr[(length - 1) / 2] + arr[(length - 1) / 2 + 1]) / 2.0;
}
return arr[length / 2];
}
时间复杂度 nlog(n)
优化方法:每次插入使用二分查找再插入的方法,代码如下:
public void addNum(int num) {
if (length == 0) {
arr[0] = num;
length ++;
return;
}
int left = 0;
int right = length - 1;
int mid = 0;
while (left <= right) {
mid = (right + left) / 2;
if (arr[mid] == num) {
break;
} else if (arr[mid] > num) {
right = mid - 1;
} else {
left = mid + 1;
}
}
if (arr[mid] > num) {
// mid--;
} else if (arr[mid] < num) {
mid ++;
}
for (int i = length; i > mid; i--) {
try {
arr[i] = arr[i - 1];
} catch (ArrayIndexOutOfBoundsException e) {
}
}
arr[mid == -1 ? 0 : mid] = num;
length ++;
}
public double findMedian() {
if (length == 0) return 0;
if (length % 2 == 0) {
return (arr[(length - 1) / 2] + arr[(length - 1) / 2 + 1]) / 2.0;
}
return arr[length / 2];
}
找到插入位置特别麻烦,边界条件太多,代码不优雅。
使用二分查找的变种,直接寻找大于等于插入值的值在数组的位置,直接插入。
关于二分查找的变形可以看这个兄弟的博客:https://www.cnblogs.com/kyoner/p/11080078.html
这个算法很难记住,但从left<right,可以判断想要跳出循环,left就要步步逼近,left++,其余的要理解。
代码如下:
/**
* 插入排序改进
* @param num
*/
public void addNum(int num) {
int left = 0;
int right = length;
int mid = 0;
// 求左边界
while (left < right) {
mid = (right + left) / 2;
if (arr[mid] == num) {
right = mid;
} else if (arr[mid] > num) {
right = mid;
} else {
left = mid + 1;
}
}
mid = left;
for (int i = length; i > mid; i--) {
arr[i] = arr[i - 1];
}
arr[mid == -1 ? 0 : mid] = num;
length ++;
}
public double findMedian() {
if (length == 0) return 0;
if (length % 2 == 0) {
return (arr[(length - 1) / 2] + arr[(length - 1) / 2 + 1]) / 2.0;
}
return arr[length / 2];
}
这样写清晰多了
还有另一种解法:使用Java优先级队列(直接使用堆排序也可解)
PriorityQueue<Integer> maxQueue;
PriorityQueue<Integer> minQueue;
int count;
/** initialize your data structure here. */
public MedianFinder() {
maxQueue = new PriorityQueue<>((x, y) -> y - x);
minQueue = new PriorityQueue<>();
count = 0;
}
public void addNum(int num) {
count ++;
maxQueue.add(num);
minQueue.add(maxQueue.poll());
if ((count & 1) != 0) {
maxQueue.add(minQueue.poll());
}
}
public double findMedian() {
if ((count & 1) == 0) {
return (maxQueue.peek() + minQueue.peek()) / 2.0;
} else {
return maxQueue.peek();
}
}
over!!!