python实现OneR算法

1.算法原理

   算法首先遍历每个特征的每一个取值，对于每一个特征值，统计它在各个类别中的出现次数，找到它出现次数最多的类别，并统计它在其他类别中的出现次数。
   举例来说，假如数据集的某一个特征可以取0或1两个值。数据集共有三个类别。特征值为0的情况下，A类有20个这样的个体，B类有60个，C类也有20个。那么特征值为0的个体最可能属于B类，当然还有40个个体确实是特征值为0，但是它们不属于B类。将特征值为0的个体分到B类的错误率就是40%，因为有40个这样的个体分别属于A类和C类。特征值为1时，计算方法类似，不再赘述；其他各特征值最可能属于的类别及错误率的计算方法也一样。
   统计完所有的特征值及其在每个类别的出现次数后，我们再来计算每个特征的错误率。计算方法为把它的各个取值的错误率相加，选取错误率最低的特征作为唯一的分类准则（OneR），用于接下来的分类。

2.源码实现

from collections import defaultdict
from operator import itemgetter
import numpy as np
from sklearn.datasets import load_iris
from sklearn.model_selection import train_test_split

dataset = load_iris()

X = dataset.data
y = dataset.target

attribute_means=X.mean(axis=0)#按列求均值

#将特征值小于均值的赋值为0，大于均值的赋值为1
X_d=np.array(X >= attribute_means,dtype='int')

def train_feature_value(X, y_true, feature_index, value):
    class_counts = defaultdict(int)

    for sample, y in zip(X, y_true):
        if sample[feature_index] == value:
            class_counts[y] += 1

    sorted_class_counts = sorted(class_counts.items(), key=itemgetter(1), reverse=True)
    most_frequent_class = sorted_class_counts[0][0]

    incorrect_predictions = [class_count for class_value, class_count in class_counts.items() if class_value != most_frequent_class]

    error = sum(incorrect_predictions)

    return most_frequent_class, error

def train_on_feature(X, y_true, feature_index):
    values = set(X[:, feature_index])

    predictors = {}
    errors = []

    for current_value in values:
        most_frequent_class, error = train_feature_value(X, y_true, feature_index, current_value)
        predictors[current_value] = most_frequent_class
        errors.append(error)

    total_error = sum(errors)

    return predictors, total_error

#random_state为随机参数，固定为某一个值时，每次执行会得到相同的结果
Xd_train, Xd_test, y_train, y_test = train_test_split(X_d, y, random_state=None)

all_predictors = {}
errors = {}

for feature_index in range(Xd_train.shape[1]):
    #print(feature_index)
    predictors, total_error = train_on_feature(Xd_train, y_train, feature_index)
    all_predictors[feature_index] = predictors
    errors[feature_index] = total_error

#print(all_predictors,errors)
best_feature, best_error = sorted(errors.items(),key=itemgetter(1))[0]
#print(best_feature,best_error)
model = {'feature':best_feature,'predictor':all_predictors[best_feature]}
#print(model)

def predict(X_test,model):
    variable = model['feature']
    #print(variable)
    predictor = model['predictor']
    #print(predictor)
    y_predicted = np.array([predictor[int(sample[variable])] for sample in X_test])

    return y_predicted

y_predicted = predict(Xd_test,model)

#计算准确率
accuracy = np.mean(y_predicted==y_test)*100

print("The test accuracy is {:.1f}%".format(accuracy))

3.运行及其结果

$ python3 example.py
The test accuracy is 63.2%