knn手写数字识别 java,利用sklearn KNN实现手写数字识别

利用sklearn KNN实现手写数字识别

```

import numpy as np

import pandas as pd

from PIL import Image

import matplotlib.pyplot as plt

from sklearn.preprocessing import MinMaxScaler

from sklearn.neighbors import KNeighborsClassifier

from sklearn.model_selection import train_test_split,GridSearchCV

```

```

def imgTovector(filename):

img=plt.imread(filename)

img_data=np.array(img)

#    Minmax=MinMaxScaler()

#    img_union=Minmax.fit_transform(img_data)

img_vector=img_data.reshape(1,-1)

return img_vector

```

```

from os import listdir

def collectFiledataset():

img_labels = list()

# 样本数据文件列表

FileList = listdir('knn_num_data')

f = len(FileList)

# 设置初始矩阵

c = 28 * 28

dataSet = np.zeros((5000, c))

for i in range(f):

path = 'knn_num_data/' + FileList[i]

trainingFilelist = listdir(path)

t = len(trainingFilelist)

for j in range(t):

fileNameStr = trainingFilelist[j]

fileStr = fileNameStr.split('.')[0]

classNum = int(fileStr.split('_')[0])

img_labels.append(classNum)

filename = './knn_num_data/' + FileList[i] + '/' + fileNameStr

img_vector = imgTovector(filename)

# 将样本数据存入矩阵

ins = i * (t - 1) + j

dataSet[ins, :] = img_vector[0]

dataSet = pd.DataFrame(dataSet)

dataSet['img_labels'] = img_labels

return dataSet

```

```

dataSet=collectFiledataset()

df=pd.DataFrame(dataSet)

df_r=df.iloc[:,:-1].std(axis=1) !=0

df_c=df.iloc[:,:-1].std(axis=0) !=0

l=df.img_labels

labels=l.loc[df_r.values]

img_df=df.loc[df_r.values,df_c.values]

img_ = img_df.loc[:,img_df.std()!=0]

#归一化

MM=MinMaxScaler()

img_union=MM.fit_transform(img_)

img_union.shape

X_train,X_test,y_train,y_test=train_test_split(feature,labels,test_size=.3)

kd_tree=KNeighborsClassifier(algorithm='kd_tree',n_neighbors=12,p=2,n_jobs=2)

# param_grid = dict(n_neighbors=np.arange(5,100))

# gc = GridSearchCV(kd_tree,param_grid,cv=4,n_jobs=2).fit(feature,labels)

```

```

kd_tree.fit(X_train,y_train)

kd_tree.score(X_train,y_train),kd_tree.score(X_test,y_test)

```