kaggle练习——手写体识别-优快云博客

本文链接：https://blog.youkuaiyun.com/zhangsh87/article/details/80339123

本文介绍了一种使用K近邻算法(KNN)进行手写数字识别的方法。通过加载MNIST数据集中的训练和测试数据，实现了从数据预处理到模型训练及预测的全过程。该文还详细展示了如何将彩色图像转换为黑白图像并进行归一化处理，最终通过KNN算法成功对手写数字进行了分类。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

# -*- coding: utf-8 -*-
"""
Created on Sun Apr 22 10:25:14 2018

@author: zhangsh
"""

import csv
import numpy as np
from sklearn.neighbors import KNeighborsClassifier

# list 转化为 array数组
def listToArray(dataList):
    m = len(dataList)  # 获取 list 的长度
    dataArray = np.zeros((1,m))
    for i in range(m):
        dataArray[0,i] = int(dataList[i])
    return dataArray

# 将彩色图片转换为黑白，归一化。将非0灰度值转化为1
def dataNormalize(data):
    m, n = data.shape
    for i in range(m):
        for j in range(n):
            if data[i,j] != 0: 
                data[i,j] = 1
    return data

# 导入训练集
def loadTrainDataSet():
    trainData = []
    trainLabel = []
    index = 0
    
    with open('train.csv','r') as file:
        readCSV = csv.reader(file)
        
        for line in readCSV:
            if index == 0 :  # 去掉第一行，第一行为说明行
                n = len(line)  # 一行的长度
                index += 1
                continue
            index += 1
            trainLabel.append(line[0])
            trainData.extend(line[1:])
        file.close()
        
    index = index - 1
    n = n - 1
    trainLabelArray = listToArray(trainLabel).reshape((index,1))
    trainDataArray = listToArray(trainData).reshape((index,n))
    
    return dataNormalize(trainDataArray), trainLabelArray

# 导入测试集
def loadTestDataSet():
    testData = []
    index = 0
    
    with open('test.csv','r') as file:
        readCSV = csv.reader(file)
        
        for line in readCSV:
            if index == 0:  # 去掉第一行
                n = len(line)  
                index += 1
                continue
            index += 1
            testData.extend(line[:])
        file.close()
        
    index = index - 1 
    testDataArray = listToArray(testData).reshape((index,n))
    return dataNormalize(testDataArray)

# 建立模型训练，并测试
def knnClassifier():
    trainingData, trainingLabel = loadTrainDataSet()  # 加载训练集
    testingData = loadTestDataSet()  # 加载测试集
    testResult = []  # 建立一个列表保存测试结果
    print(testingData.shape)
    knn = KNeighborsClassifier(algorithm = 'ball_tree')  # 建立KNN模型
    knn.fit(trainingData, trainingLabel)  # 训练模型
    
    testResult.append(('ImageId','Label')) 

    i = 1
    for line in testingData:  
        predictLabel = knn.predict(line.reshape((1,-1)))
        testResult.append((i,int(predictLabel[0])))  
        print('预测第%d条数据' %i)
        i+=1
        
    with open('result.csv','w',newline='') as file:  
        writer=csv.writer(file)
        writer.writerows(testResult)
if __name__ == "__main__":
    knnClassifier()

参考了这篇博客： https://blog.youkuaiyun.com/u012198382/article/details/64907162