Python实验--手写五折交叉验证+调库实现SVM/RFC/KNN手写数字识别

1. 数据读取

先说一下要用到的数据集：

数据集自取地址：

链接：https://pan.baidu.com/s/1Vd2ADHEalSNnuOEcPJD8gQ
提取码：3hk6

数据集构成：

0-9十个数字，总共1934个样本，以数字_n命名，每个样本为32*32大小的txt文件（事先将图片处理后二值化）

数据读取代码：

def img2vector(filename):
    # 创建向量
    returnVect = np.zeros((1, 1024))
    # 打开数据文件,读取每行内容
    fr = open(filename)
    for i in range(32):
        # 读取每一行
        lineStr = fr.readline()
        # 将每行前32字符转成int,存入向量
        for j in range(32):
            returnVect[0, 32 * i + j] = int(lineStr[j])
    return returnVect

最后生成数据集形式：

X：1934*1024的矩阵，每行代表一个样本，1024列为每个样本的1024个像素点（二值表示）

Y：1934*1的矩阵，每一行对应相同下标X中样本的标签

数据集生成代码：

def trainData(trainPath):
    trainfile = os.listdir(trainPath)  # 获取训练集文件下的所有文件名
    Y = np.zeros((len(trainfile), 1))
    # 先建立一个行数为训练样本数。列数为1024的0数组矩阵，1024为图片像素总和，即32*32
    X = np.zeros((len(trainfile), 1024))
    # 取文件名的第一个数字为标签名
    for i in range(0, len(trainfile)):
        thislabel = trainfile[i].split(".")[0].split("_")[0]
        if len(thislabel) != 0:
            Y[i][0] = int(thislabel)  # 保存标签
        X[i, :] = img2vector(trainPath + "/" + trainfile[i])  # 将训练数据写入0矩阵
    return X, Y

2. K折交叉验证

关于K折交叉验证的原理其他博文都有详细说明，我这就不赘述了，直接上代码

（本人水平有限，实现方法繁琐，但是能用，手动狗头保命）

代码思路介绍：

1. 将每个数字对应的下标打乱

2. 由于每个数字样本数不一致，这里将其统一到200，不足的随机抽样本补充，多余的随机剔除

3. 按照打乱后的下标生成shuffle后的数据集

4. 按照每个数字40个样本的方式提取出五折子数据集

详细代码：

# import dataset
X, Y = trainData('data1')
size = [0, 0, 0, 0, 0, 0, 0, 0, 0, 0] # 保存每个数字的样本数
for i in range(1934):
    size[int(Y[i][0])] += 1
# shuffle and balance
position = []
for i in range(10):
    left = 0
    right = 0
    for k in range(i + 1):
        right += size[k]
    left = right - size[i] # 计算每个数字对应的下标左右边界
    ran = list(range(left, right))
    random.shuffle(ran) # 生成打乱后的下标列表
    if len(ran) < 200:
        for j in range(200 - len(ran)):
            ran.append(np.random.randint(left, right)) # 随机抽取补充
    if len(ran) > 200:
        for j in range(len(ran) - 200):
            del ran[-1] # 随机剔除
    position.append(ran)
X_shuffled = np.zeros((2000, 1024))
for i in range(10):
    for j in range(200):
        x = X[position[i][j]]
        X_shuffled[j + 200*i] = x # 按照打乱后的下标生成新数据集
# split into 5 parts
X_part = []
for i in range(5):
    X_split = np.zeros((400, 1024))
    for j in range(10):
        for k in range(40):
            X_split[(k + 40*j), :] = X_shuffled[(k + 200*j + 40*i), :] 
    X_part.append(X_split)
Y_part = []
for i in range(10):
    for j in range(40):
        Y_part.append(i) #生成对应的标签集

3. 调库实现手写数字识别

1. RFC

# K-Folder
score =[]
for i in range(5):
    X_test = X_part[i]
    Y_test = Y_part
    X_train = np.concatenate((X_part[(i+1) % 5], X_part[(i+2) % 5], X_part[(i+3) % 5], X_part[(i+4) % 5]), axis=0)
    Y_train = Y_test*4
    clf = RandomForestClassifier(n_estimators=200, criterion='gini', max_depth=None,
                                 min_samples_split=3, min_samples_leaf=1, min_weight_fraction_leaf=0.0,
                                 max_features='auto', max_leaf_nodes=None, min_impurity_decrease=0.0,
                                 min_impurity_split=None, bootstrap=True, oob_score=False, n_jobs=1,
                                 random_state=None, verbose=0, warm_start=False, class_weight=None)
    # train
    clf.fit(X_train, Y_train)
    score.append(clf.score(X_test, Y_test))
    Y_pred = clf.predict(X_test)
    cm = confusion_matrix(Y_test, Y_pred)
    plt.matshow(cm)
    plt.title('epoch %d' % (i + 1))
plt.show()
acc_sum = 0
for i in range(5):
    acc_sum += score[i]
print("Average Acc: %f" % (acc_sum / 5))

1. SVM

score =[]
for i in range(5):
    X_test = X_part[i]
    Y_test = Y_part
    X_train = np.concatenate((X_part[(i+1) % 5], X_part[(i+2) % 5], X_part[(i+3) % 5], X_part[(i+4) % 5]), axis=0)
    Y_train = Y_test*4
    clf = svm.SVC(C=200.0, kernel='rbf', degree=3, gamma='auto',
                  coef0=0.0, shrinking=True, probability=False, tol=0.001,
                  cache_size=200, class_weight=None, verbose=False,
                  max_iter=-1, decision_function_shape='ovr',
                  random_state=None)
    # train
    clf.fit(X_train, Y_train)
    score.append(clf.score(X_test, Y_test))
    Y_pred = clf.predict(X_test)
    cm = confusion_matrix(Y_test, Y_pred)
    plt.matshow(cm)
    plt.title('epoch %d' % (i + 1))
plt.show()
acc_sum = 0
for i in range(5):
    acc_sum += score[i]
print("Average Acc: %f" % (acc_sum / 5))

1. KNN

score =[]
plt.figure()
for i in range(5):
    X_test = X_part[i]
    Y_test = Y_part
    X_train = np.concatenate((X_part[(i+1) % 5], X_part[(i+2) % 5], X_part[(i+3) % 5], X_part[(i+4) % 5]), axis=0)
    Y_train = Y_test*4
    clf = KNeighborsClassifier(n_neighbors=3, weights='uniform',
                               algorithm='auto', leaf_size=30,
                               p=2, metric='minkowski', metric_params=None,
                               n_jobs=None, )
    # train
    clf.fit(X_train, Y_train)
    score.append(clf.score(X_test, Y_test))
    Y_pred = clf.predict(X_test)
    cm = confusion_matrix(Y_test, Y_pred)
    plt.matshow(cm)
    plt.title('epoch %d' % (i+1))
plt.show()
acc_sum = 0
for i in range(5):
    acc_sum += score[i]
print("Average Acc: %f" % (acc_sum / 5))

4. 用到的库列表

# KNN
import random
import numpy as np
import os
from sklearn.neighbors import KNeighborsClassifier
import matplotlib.pyplot as plt
from sklearn.metrics import confusion_matrix
# SVM
import random
import numpy as np
import os
import matplotlib.pyplot as plt
from sklearn.metrics import confusion_matrix
from sklearn import svm
# RFC
import random
import numpy as np
import os
from sklearn. ensemble import RandomForestClassifier
import matplotlib.pyplot as plt
from sklearn.metrics import confusion_matrix