ACO-SVM手语识别系统报错：X含63特征，SVC预期31输入特征

报错‘X has 63 features, but SVC is expecting 31 features as input’的核心问题很明确：训练SVM时用了ACO筛选出的31个特征，但预测时直接传入了完整的63维测试数据，两边特征维度对不上。

修复步骤 #

1. 保存ACO选中的特征索引
   在HybridACOSVM类中新增一个属性，专门存储训练阶段ACO选出的特征索引，确保预测时能复用同一组特征。
2. 修改预测逻辑
   调用predict方法时，先对测试数据做和训练阶段一样的特征筛选，再传入SVM模型进行预测。

修改后的完整代码 #

# Load the dataset
import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
from sklearn.svm import SVC

df = pd.read_csv('dataset.csv')
df.columns = [i for i in range(df.shape[1])]

# Split the data into features and labels
X = df.iloc[:, :-1]
Y = df.iloc[:, -1]

# Split the data into training and testing sets
x_train, x_test, y_train, y_test = train_test_split(X, Y, test_size=0.2, random_state=0)

# Apply feature scaling to the training and testing data
scaler = StandardScaler()
x_train_scaled = scaler.fit_transform(x_train)
x_test_scaled = scaler.transform(x_test)

# Define the ACO-SVM model
class HybridACOSVM:
    def __init__(self, svm_params, aco_params):
        self.svm_params = svm_params
        self.aco_params = aco_params
        self.svm = None
        self.selected_features = None  # 新增：存储选中的特征索引

    def train(self, x_train, y_train):
        # Perform ACO feature selection
        selected_features = self.perform_aco(x_train, y_train)
        self.selected_features = selected_features  # 保存选中的特征

        # Train SVM using selected features
        self.svm = SVC(**self.svm_params)
        self.svm.fit(x_train[:, selected_features], y_train)

    def perform_aco(self, x_train, y_train):
        # Perform ACO feature selection algorithm
        # Your ACO algorithm implementation here

        # Randomly select some features as a placeholder
        num_features = x_train.shape[1]
        num_selected = int(num_features * 0.5)  # Select half of the features
        selected_features = np.random.choice(num_features, num_selected, replace=False)

        return selected_features

    def predict(self, x_test):
        # Make predictions using the trained SVM model
        # 新增：先筛选特征，再预测
        x_test_selected = x_test[:, self.selected_features]
        return self.svm.predict(x_test_selected)

# Set the parameters for SVM and ACO
svm_params = {
    'C': 10,
    'gamma': 0.1,
    'kernel': 'rbf'
}

aco_params = {
    # Parameters for the ACO algorithm
    # Adjust these parameters according to your implementation
}

# Train the hybrid ACO-SVM model
hybrid_model = HybridACOSVM(svm_params, aco_params)
hybrid_model.train(x_train_scaled, y_train)

# Make predictions on the test set
y_pred = hybrid_model.predict(x_test_scaled)

关键修改点标注 #

• 在HybridACOSVM的__init__方法中添加self.selected_features = None
• 在train方法中把ACO选出的selected_features赋值给self.selected_features
• 在predict方法中先对测试数据做特征筛选x_test_selected = x_test[:, self.selected_features]，再传入SVM预测

内容的提问来源于stack exchange，提问作者Iorolun Emmanuel