#include \n#include \n#include \n\n// \u5B9A\u4E49\u51B3\u7B56\u6811\u8282\u70B9\u7ED3\u6784\nstruct Node {\n int featureIndex; \n double threshold; \n int label; \n Node* left; \n Node* right; \n};\n\n// \u968F\u673A\u6811\u6811\u7C7B\u522B\u5668\u7C7B\nclass RandomForestClassifier {\npublic:\n RandomForestClassifier(int numTrees, int maxDepth) : numTrees(numTrees), maxDepth(maxDepth) {}\n\n void fit(const std::vector<std::vector>& X, const std::vector& y) {\n // \u521B\u5EFA\u5E76\u8BAD\u7EC3\u968F\u673A\u6811\u6811\u5185\u7684\u51B3\u7B56\u6811\n for (int i = 0; i < numTrees; ++i) {\n std::vector<std::vector> bootstrapX; \n std::vector bootstrapY; \n generateBootstrapSample(X, y, bootstrapX, bootstrapY); \n\n Node* root = buildTree(bootstrapX, bootstrapY, maxDepth); \n trees.push_back(root); \n } \n } \n\n int predict(const std::vector& sample) {\n // \u5BF9\u6BCF\u682A\u51B3\u7B56\u6811\u8FDB\u884C\u9884\u6D4B\uFF0C\u5E76\u4F7F\u7528\u6295\u7968\u6CD5\u786E\u5B9A\u6700\u7EC8\u5206\u7C7B\n std::vector predictions(numTrees, 0); \n for (int i = 0; i < numTrees; ++i) {\n predictions[i] = traverseTree(sample, trees[i]); \n } \n\n std::vector counts(2, 0); \n for (int i = 0; i < numTrees; ++i) {\n counts[predictions[i]]++; \n } \n\n return (counts[0] > counts[1]) ? 0 : 1; \n } \n\nprivate:\n int numTrees; \n int maxDepth; \n std::vector<Node*> trees; \n\n // \u751F\u6210\u590D\u6709\u6837\u672C\n void generateBootstrapSample(const std::vector<std::vector>& X, const std::vector& y, \n std::vector<std::vector>& bootstrapX, std::vector& bootstrapY) {\n std::random_device rd; \n std::mt19937 gen(rd()); \n std::uniform_int_distribution dist(0, X.size() - 1); \n\n for (int i = 0; i < X.size(); ++i) {\n int index = dist(gen); \n bootstrapX.push_back(X[index]); \n bootstrapY.push_back(y[index]); \n } \n } \n\n // \u6784\u5EFA\u51B3\u7B56\u6811\n Node* buildTree(const std::vector<std::vector>& X, const std::vector& y, int depth) {\n if (depth == 0 || y.empty()) \n return nullptr; \n\n int numFeatures = X[0].size(); \n std::random_device rd; \n std::mt19937 gen(rd()); \n std::uniform_int_distribution dist(0, numFeatures - 1); \n\n int featureIndex = dist(gen); \n double threshold = findBestThreshold(X, y, featureIndex); \n\n std::vector<std::vector> leftX, rightX; \n std::vector leftY, rightY; \n for (int i = 0; i < X.size(); ++i) {\n if (X[i][featureIndex] < threshold) {\n leftX.push_back(X[i]); \n leftY.push_back(y[i]); \n } else {\n rightX.push_back(X[i]); \n rightY.push_back(y[i]); \n } \n } \n\n Node* node = new Node; \n node->featureIndex = featureIndex; \n node->threshold = threshold; \n node->left = buildTree(leftX, leftY, depth - 1); \n node->right = buildTree(rightX, rightY, depth - 1); \n\n if (leftY.size() > rightY.size()) {\n node->label = majorityVote(leftY); \n } else {\n node->label = majorityVote(rightY); \n } \n\n return node; \n } \n\n // \u627E\u5230\u6700\u4F73\u9600\u503C\n double findBestThreshold(const std::vector<std::vector>& X, const std::vector& y, int featureIndex) {\n double bestThreshold = 0.0; \n double bestGini = 1.0; \n\n std::vector featureValues; \n for (int i = 0; i < X.size(); ++i) {\n featureValues.push_back(X[i][featureIndex]); \n } \n\n std::sort(featureValues.begin(), featureValues.end()); \n\n for (int i = 0; i < featureValues.size() - 1; ++i) {\n double threshold = (featureValues[i] + featureValues[i + 1]) / 2.0; \n\n std::vector leftY, rightY; \n for (int j = 0; j < X.size(); ++j) {\n if (X[j][featureIndex] < threshold) {\n leftY.push_back(y[j]); \n } else {\n rightY.push_back(y[j]); \n } \n } \n\n double gini = calculateGini(leftY, rightY); \n if (gini < bestGini) {\n bestGini = gini; \n bestThreshold = threshold; \n } \n } \n\n return bestThreshold; \n } \n\n // \u8BA1\u7B97\u57FA\u5C04\u6307\u6570\n double calculateGini(const std::vector& leftY, const std::vector& rightY) {\n int totalSize = leftY.size() + rightY.size(); \n double leftGini = calculateImpurity(leftY) * leftY.size() / totalSize; \n double rightGini = calculateImpurity(rightY) * rightY.size() / totalSize; \n\n return leftGini + rightGini; \n } \n\n // \u8BA1\u7B97\u4E0D\u7B26\u5EFA\u5EA6\n double calculateImpurity(const std::vector& y) {\n if (y.empty()) \n return 0.0; \n\n int count0 = 0, count1 = 0; \n for (int i = 0; i < y.size(); ++i) {\n if (y[i] == 0) \n count0++; \n else \n count1++; \n } \n\n double p0 = static_cast(count0) / y.size(); \n double p1 = static_cast(count1) / y.size(); \n\n return 1.0 - (p0 * p0 + p1 * p1); \n } \n\n // \u4F7F\u7528\u51B3\u7B56\u6811\u8FDB\u884C\u9884\u6D4B\n int traverseTree(const std::vector& sample, Node* root) {\n if (root == nullptr) \n return -1; \n\n if (root->left == nullptr && root->right == nullptr) \n return root->label; \n\n if (sample[root->featureIndex] < root->threshold) \n return traverseTree(sample, root->left); \n else \n return traverseTree(sample, root->right); \n } \n\n // \u6295\u7968\u6CD5\u786E\u5B9A\u6700\u7EC8\u5206\u7C7B\n int majorityVote(const std::vector& predictions) {\n int count0 = 0, count1 = 0; \n for (int i = 0; i < predictions.size(); ++i) {\n if (predictions[i] == 0) \n count0++; \n else \n count1++; \n } \n\n return (count0 > count1) ? 0 : 1; \n } \n};\n\nint main() {\n std::vector<std::vector> X = {{0.5, 0.2}, {0.3, 0.8}, {0.8, 0.6}, {0.6, 0.1}}; \n std::vector y = {0, 1, 1, 0}; \n\n RandomForestClassifier rfClassifier(3, 2); \n rfClassifier.fit(X, y); \n\n std::vector sample = {0.7, 0.3}; \n int prediction = rfClassifier.predict(sample); \n\n std::cout << "Prediction: " << prediction << std::endl; \n\n return 0; \n} \n\n\u8FD9\u662F\u4E00\u4E2A\u7B80\u5355\u7684\u968F\u673A\u6811\u6811\u7C7B\u522B\u5668\u7684\u5B9E\u73B0\u3002\u5B83\u5305\u542B\u4E86\u51B3\u7B56\u6811\u8282\u70B9\u7ED3\u6784 Node \u548C\u968F\u673A\u6811\u6811\u7C7B\u522B\u5668\u7C7B RandomForestClassifier\u3002\u5728 main \u51FD\u6570\u4E2D\uFF0C\u6211\u4EEC\u4F7F\u7528\u793A\u4F8B\u6570\u636E\u8FDB\u884C\u8BAD\u7EC3\uFF0C\u5E76\u5BF9\u4E00\u4E2A\u65B0\u6837\u672C\u8FDB\u884C\u9884\u6D4B\u3002\n\n\u8BF7\u6CE8\u610F\uFF0C\u8FD9\u53EA\u662F\u4E00\u4E2A\u7B80\u5355\u7684\u793A\u4F8B\uFF0C\u5B9E\u9645\u5E94\u7528\u4E2D\u53EF\u80FD\u9700\u8981\u8FDB\u4E00\u6B65\u4F18\u5316\u548C\u6539\u8FDB\u3002
