【大数据】孤立森林算法 - ToB企服应用市场:ToB评测及商务社交产业平台

from sklearn.ensemble import IsolationForest
# 假设我们有一些数据
data = [[-1.1], [0.3], [0.8], [1.1], [10.0]]
# 创建一个孤立森林模型
model = IsolationForest(max_samples=100, random_state=42)
# 使用模型来训练数据
model.fit(data)
# 预测每个点是否异常
anomaly_scores = model.decision_function(data)
print("Anomaly scores:", anomaly_scores)
# 找出异常的点
is_anomaly = model.predict(data)
print("Is anomaly?", is_anomaly)

复制代码

import java.util.Arrays;
import java.util.Random;
public class IsolationForest {
private int nFeatures;
private int nTrees;
private int maxDepth;
private double contamination;
public IsolationForest(int nFeatures, int nTrees, int maxDepth, double contamination) {
this.nFeatures = nFeatures;
this.nTrees = nTrees;
this.maxDepth = maxDepth;
this.contamination = contamination;
}
public double[] run(double[][] data) {
// 生成随机森林
Random random = new Random();
for (int i = 0; i < nTrees; i++) {
// 构建随机子集
int[] indices = new int[data.length];
for (int j = 0; j < indices.length; j++) {
indices[j] = j;
}
random.nextInt();
for (int j = 0; j < data.length; j++) {
int index = j + random.nextInt(data.length - j);
int temp = indices[index];
indices[index] = indices[j];
indices[j] = temp;
}
// 选择随机特征
int[] features = new int[nFeatures];
for (int j = 0; j < nFeatures; j++) {
features[j] = j;
}
for (int j = 0; j < nFeatures; j++) {
int index = j + random.nextInt(nFeatures - j);
int temp = features[index];
features[index] = features[j];
features[j] = temp;
}
// 构建决策树
// ...
}
// 计算每个点的异常得分
double[] scores = new double[data.length];
for (int i = 0; i < data.length; i++) {
// 计算每棵树的异常得分之和
// ...
scores[i] = ...;
}
// 计算异常概率
double threshold = getThreshold(scores);
for (int i = 0; i < scores.length; i++) {
if (scores[i] < threshold) {
scores[i] = 1.0;
} else {
scores[i] = 0.0;
}
}
return scores;
}
private double getThreshold(double[] scores) {
Arrays.sort(scores);
int n = (int) Math.ceil(contamination * scores.length);
return scores[n - 1];
}
public static void main(String[] args) {
// 示例数据
double[][] data = {
{1, 1},
{1, 1},
{1, 1},
{2, 2},
{2, 2},
{2, 2},
{3, 3},
{3, 3},
{3, 3},
{4, 4},
{4, 4},
{4, 4},
{5, 5},
{5, 5},
{5, 5},
{6, 6},
{6, 6},
{6, 6},
{7, 7},
{7, 7},
{7, 7},
{8, 8},
{8, 8},
{8, 8}
};
// 构建孤立森林
IsolationForest iforest = new IsolationForest(2, 100, 3, 0.05);
// 运

复制代码

#include <iostream>
#include <vector>
#include <cmath>
#include <ctime>
#include <random>
using namespace std;
struct IsolationTree {
int treeSize;
vector<pair<double, double>> samples;
IsolationTree(int size, vector<pair<double, double>> data) : treeSize(size), samples(data) {}
void buildTree() {
// 随机选择一个点作为树的根节点
int rootIndex = rand() % samples.size();
double rootValue = samples[rootIndex].first;
double rootRange = sqrt(samples.size()) * 2;
// 构建树的过程
// ...
}
void searchAnomalies(double threshold) {
// 搜索树中的异常点
// ...
}
};
int main() {
srand(time(0)); // 初始化随机数生成器
vector<pair<double, double>> data = {{1, 2}, {3, 4}, {5, 6}, {7, 8}}; // 示例数据
int treeSize = 10; // 森林中每棵树的大小
double threshold = 2.5; // 异常阈值
// 构建孤立森林
vector<IsolationTree> forest;
for (int i = 0; i < 10; ++i) { // 构建10棵树作为森林
forest.emplace_back(treeSize, data);
forest.back().buildTree();
}
// 搜索异常
for (auto& tree : forest) {
tree.searchAnomalies(threshold);
}
return 0;
}

复制代码