Baseline Models

最新推荐文章于 2025-01-03 15:58:40 发布

原创最新推荐文章于 2025-01-03 15:58:40 发布 · 550 阅读

0 ·

CC 4.0 BY-SA版权

机器学习算法专栏收录该内容

6 篇文章

订阅专栏

本文介绍了两种基本的机器学习预测方法：随机预测算法和零规则算法，前者通过随机选择训练集中的输出值进行预测，后者则依据训练集中最常见的类别或平均数值进行预测。

Chapter 5

Baseline Models

随机预测算法

# Example of Making Random Predictions 
from random import seed 
from random import randrang

# Generate random predictions 
def random_algorithm(train, test): 
	output_values = [row[-1] for row in train] #[0,1,0,1,0,1]
	unique = list(set(output_values)) #[0,1]
	predicted = list() 
	for row in test: 
		index = randrange(len(unique)) #0 or 1
		predicted.append(unique[index]) 
	return predicted
	
seed(1) 
train = [[0], [1], [0], [1], [0], [1]] 
test = [[None], [None], [None], [None]] 
predictions = random_algorithm(train, test) 
print(predictions)

结果

[0,0,1,0]

零规则算法

# Example of Zero Rule Classification Predictions 
from random import seed

# zero rule algorithm for classification 
def zero_rule_algorithm_classification(train, test): 
	output_values = [row[-1] for row in train] 
	prediction = max(set(output_values), key=output_values.count) #0,返回的是数量做多的
	predicted = [prediction for i in range(len(test))] 
	return predicted
seed(1) 
train = [['0'], ['0'], ['0'], ['0'], ['1'], ['1']] 
test = [[None], [None], [None], [None]] 
predictions = zero_rule_algorithm_classification(train, test) 
print(predictions)

结果：

['0', '0', '0', '0']

回归

在这里插入图片描述

# Example of Zero Rule Regression Predictions 
from random import seed

# zero rule algorithm for regression 
def zero_rule_algorithm_regression(train, test): 
	output_values = [row[-1] for row in train] 
	prediction = sum(output_values) / float(len(output_values)) 
	predicted = [prediction for i in range(len(test))] 
return predicted

seed(1) 
train = [[10], [15], [12], [15], [18], [20]] 
test = [[None], [None], [None], [None]] 
predictions = zero_rule_algorithm_regression(train, test) 
print(predictions)