Revenge of kNN

Time Limit: 2000/1000 MS (Java/Others)    Memory Limit: 32768/32768 K (Java/Others)
Total Submission(s): 584    Accepted Submission(s): 136

Problem Description
In
pattern recognition, the k-Nearest Neighbors algorithm (or k-NN for
short) is a non-parametric method used for classification and
regression. In both cases, the input consists of the k closest training
examples in the feature space.
In k-NN regression, the output is the
property value for the object. This value is the average of the values
of its k nearest neighbors.
---Wikipedia

Today, kNN takes
revenge on you. You have to handle a kNN case in one-dimensional
coordinate system. There are N points with a position Xi and value Vi.
Then there are M kNN queries for point with index i, recalculate its
value by averaging the values its k-Nearest Neighbors. Note you have to
replace the value of i-th point with the new calculated value. And if
there is a tie while choosing k-Nearest Neighbor, choose the one with
the minimal index first.

 
Input
The first line contains a single integer T, indicating the number of test cases.

Each
test case begins with three integers N, M and K, in which K indicating
the number of k-Nearest Neighbors. Then N lines follows, each line
contains two integers Xi and Vi. Then M lines with the queried index Qi
follows.

[Technical Specification]
1. 1 <= T <= 5
2. 2<=N<= 100 000, 1<=M<=100 000
3. 1 <= K <= min(N – 1, 10)
4. 1 <= Vi <= 1 000
5. 1 <= Xi <= 1 000 000 000, and no two Xi are identical.
6. 1 <= Qi <= N

 
Output
For each test case, output sum of all queries rounded to six fractional digits.
 
Sample Input
1
5 3 2
1 2
2 3
3 6
4 8
5 8
2
3
4
 
Sample Output
17.000000
做这题让我知道了如何解决一个很常见但是也很难想的问题:一个数组如果排序后被打乱了,知道它原来的位置,如何对应出现在的位置?
题意:X轴上有 n 个点,每个点都有个初始位置和权值(这些位置是杂乱无章的),现在给m次询问:每次询问第 i 个给出的点,找出离他最近的 k 个点,然后将 k 个点的权值相加取平均值赋值给当前询问的点,输出的最终答案就是这m 次询问每次询问的平均值之和。还有就是关于 k 个点的选取,当距离相同时,选择给出顺序小的。
题解:由于要找到邻居,排序是无疑的,但是询问却是询问的以前的位置。排完序之后如何找到原来的位置??循环?不行,10^5次询问每次找点也要10^5,肯定会超时?这时我们可以利用一个数组来记录排完序后元素原来的位置。但是,元素的下标是从 1 <= Xi <= 1 000 000 000,数组肯定存不下,怎么办?离散化呗,这样寻找的时间就变成 O(1)了。
可能有人不懂离散化后数组存的意思,这里举个例子:
原来的X轴分布假设为:
3 1 2 100 99
我们记录一下每个Xi 出现的位置 3(1) 1(2) 2(3) 100(4) 99(5)
排个序: 1(2) 2(3) 3(1) 99(5) 100(4)
弄个数组将括号里面的数表示成下标分别赋值给 1-5  a[2]=1 a[3]=2 a[1]=3 a[5]=4 a[4]=5
然后对应查询,假设我们查询初始位置是第 4 位的那个 Xi=100,对应排完序后的数组的位置是 a[4] = 5 而新的数组第5个数字等于100!!是不是很神奇,这就是离散化的好处了.不懂我说的同学可以看下acdreamer大神的博客:http://blog.csdn.net/acdreamers/article/details/8520096
 
#include <iostream>
#include <stdio.h>
#include <string.h>
#include <algorithm>
using namespace std;
typedef long long LL;
const int N = ;
struct Node{
LL x;
int id1;
double v;
}node[N];
int idx[N];
int cmp(Node a,Node b){
return a.x<b.x;
}
int main()
{
int tcase,n,m,k,x;
scanf("%d",&tcase);
while(tcase--)
{
scanf("%d%d%d",&n,&m,&k);
for(int i=;i<=n;i++){
scanf("%lld%lf",&node[i].x,&node[i].v);
node[i].id1 = i;
}
sort(node+,node++n,cmp);
for(int i=;i<=n;i++){
idx[node[i].id1] = i;
}
double sum = ;
while(m--){
int id;
scanf("%d",&id);
int now = idx[id];
double avg = ;
int l = now-,r = now+;
for(int i=;i<=k;i++){
if(l>=&&r<=n){
LL dis1 = node[now].x - node[l].x;
LL dis2 = node[r].x - node[now].x;
if(dis1<dis2){
avg+=node[l--].v;
}else if(dis1>dis2){
avg+=node[r++].v;
}else { ///相等的话按照邻居原来的下标进行选择
int ori_l = node[l].id1;
int ori_r = node[r].id1;
if(ori_l<ori_r) {
avg+=node[l--].v;
}else{
avg+=node[r++].v;
}
}
}else if(l>=){
avg+=node[l--].v;
}else if(r<=n){
avg+=node[r++].v;
}
}
/* ///不知道WA的原因
int l=1,r=1;
for(int i=1;i<=k;i++){
if(now-l>=1&&now+r<=n){
LL dis1 = node[now].x - node[now-l].x;
LL dis2 = node[now+r].x - node[now].x;
if(dis1<dis2){
avg+=node[now-l].v;
l++;
}else if(dis1>dis2){
avg+=node[now+r].v;
r++;
}else { ///相等的话按照邻居原来的下标进行选择
int ori_l = node[now+l].id1;
int ori_r = node[now+r].id1;
if(ori_l<ori_r) {
avg+=node[now-l].v;
l++;
}else{
avg+=node[now+r].v;
r++;
}
}
}else if(now-l>=1){
avg+=node[now-l].v;
l++;
}else if(now+r<=n){
avg+=node[now+r].v;
r++;
}
}*/
node[now].v = avg/k;
sum+=avg/k;
}
printf("%.6lf\n",sum);
}
return ;
}

最新文章

  1. 深入理解JSONP原理——前端面试
  2. .NET涉及的一些名词
  3. iOS-SDWebImage
  4. SQL Server 2008 评估期已过解决方法
  5. ServiceStack
  6. css3 text-transform变形动画
  7. WordPress社会化评论插件多说、有言、灯鹭、评论啦
  8. PHP中$_GET[&#39;name&#39;]与$_POST[&#39;name&#39;]变量直接用变量名$name的php配置
  9. Swift学习笔记(一)搭配环境以及代码运行成功
  10. 在hive中直接对timestamp类型取max报错
  11. dirlock.go
  12. .net 使用oracle 的存储过程有返回值也有数据集(游标)
  13. LeetCode(65):有效数字
  14. HUAWEI USG6000系列 &amp; NGFW Module V100R001 典型配置案例
  15. TFS 报错解决方案:tf400324
  16. k8s weave network IP回收冲突
  17. ArcGIS案例学习笔记2_1_山顶点提取最大值提取
  18. iOS常用代码总结
  19. 如何获取 docker 容器(container)的 ip 地址
  20. Spring Boot(十二):Spring Boot 如何测试打包部署

热门文章

  1. 单片机入门学习笔记5:STC下载器
  2. BFS:HDU3085-Nightmare Ⅱ(双向BFS)
  3. 【高精度】模板 (C++)
  4. hibernate实体xml一对多关系映射
  5. ECMAScript5.1
  6. uReplicator实现分析
  7. CentOS安装jdk和tomcat
  8. caffe工程配置问题
  9. 初识面向对象-python
  10. 一个iOS程序员眼中的跨域问题