整理digit-recognizer几种解决方案

最新推荐文章于 2025-05-30 09:00:02 发布

Gunther17

最新推荐文章于 2025-05-30 09:00:02 发布

阅读量872

点赞数 1

CC 4.0 BY-SA版权

分类专栏： kaggle笔记

本文链接：https://blog.youkuaiyun.com/dongyanwen6036/article/details/81542900

kaggle笔记专栏收录该内容

2 篇文章

订阅专栏

本文介绍了在Windows 10环境下使用Anaconda解决TensorFlow导入错误的问题，详细阐述了从Python 2.7升级到Python 3.5以及如何通过conda命令安装TensorFlow CPU版本。此外，还探讨了digit-recognizer问题的随机森林方法，并展示了CNN模型的初步应用和训练结果，尽管数据量大导致训练时间较长。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

先放上理想曲线：
这里写图片描述

几种方法代码：

#!usr/bin/python
#codeing: utf-8
'''
Create on 2018-08-09
Author: Gunther17
   Ctrl + 1: 注释/反注释

　　Ctrl + 4/5: 块注释/块反注释

　　Ctrl + L: 跳转到行号

　　Tab/Shift + Tab: 代码缩进/反缩进

　　Ctrl ＋I：显示帮助
'''
import os.path
import csv
import time
import numpy as np
import pandas as pd
from sklearn.decomposition import PCA
from sklearn.neighbors import KNeighborsClassifier
from sklearn.metrics import accuracy_score
from sklearn.tree import DecisionTreeClassifier
from sklearn.ensemble import RandomForestClassifier
from sklearn.svm import SVC


data_dir='D:\data Competition\digit-recognizer/'

#加载数据
def opencsv():
    data_train=pd.read_csv(os.path.join(data_dir,'input/train.csv'))
    data_test=pd.read_csv(os.path.join(data_dir,'input/test.csv'))

    train_data=data_train.values[0:,1:] #读入全部训练数据,  [行，列]
    train_label=data_train.values[0:,0] # 读取列表的第一列
    test_data=data_test.values[0:,0:]  # 测试全部测试个数据

    return train_data,train_label,test_data


def saveResults(result,csvName):
    with open(csvName,'w') as myfile:
        '''
       创建记录输出结果的文件（w 和 wb 使用的时候有问题）
       python3里面对 str和bytes类型做了严格的区分，不像python2里面某些函数里可以混用。
       所以用python3来写wirterow时，打开文件不要用wb模式，只需要使用w模式，然后带上newline=''
       '''
        mywrite=csv.writer(myfile)
        mywrite.writerow(["ImageId","Label"])
        index=0
        for r in result:
            index+=1
            mywrite.writerow([index,int(r)])
        print('Saved successfully....')


def knnClassify(traindata,trainlabel):
    print('Train knn...')
    knnClf=KNeighborsClassifier() # default:k = 5,defined by yourself:KNeighborsClassifier(n_neighbors=10)
    knnClf.fit(traindata,np.ravel(trainlabel))# ravel/flattened 返回1维 array，其中flatten函数返回的是拷贝。.
    return knnClf

def dtClassify(traindata,trainlabel):
    print('Train decision tree...')
    dtClf=DecisionTreeClassifier()
    dtClf.fit(traindata,np.ravel(trainlabel))
    return dtClf

def rfClassify(traindata,trainlabel):
    print('Train Random forest...')
    rfClf=RandomForestClassifier()
    rfClf.fit(traindata,np.ravel(trainlabel))
    return rfClf

def svmClassify(traindata,trainlabel):
    print('Train svm...')
    svmClf=SVC(C=4,kernel='rbf')
    svmClf.fit(traindata,np.ravel(trainlabel))
    return svmClf



def dpPCA(x_train,x_test,Com):
    print('dimension reduction....')
    trainData=np.array(x_train)
    testData=np.array(x_test)
    '''
    n_components>=1
      n_components=NUM   设置占特征数量比
    0 < n_components < 1
      n_components=0.99  设置阈值总方差占比
    '''
    pca=PCA(n_components=Com,whiten=False)
    pca.fit(trainData) #fit the model with X
    pcaTrainData=pca.transform(trainData)# 在 X上进行降维
    pcaTestData=pca.transform(testData)
    #pca 方差大小 方差占比 特征数量
   # print(pca.explained_variance_,'\n',pca.explained_variance_ratio_,'\n',pca.components_)
    return pcaTrainData,pcaTestData

def dRecognition_knn():
    start_time=time.time()
    #load data
    trainData,trainLabel,testData=opencsv()
    print('load data finish...')
    stop_time1=time.time()
    print('load data take: %f' %(stop_time1-start_time))

    #dimension reduction
    trainData,testData=dpPCA(trainData,testData,0.8)

    knnClf=knnClassify(trainData,trainLabel)
    dtClf=dtClassify(trainData,trainLabel)
    rfClf=rfClassify(trainData,trainLabel)
    svmClf=svmClassify(trainData,trainLabel)


    trainlabel_knn=knnClf.predict(trainData)
    trainlabel_dt=dtClf.predict(trainData)
    trainlabel_rf=rfClf.predict(trainData)
    trainlabel_svm=svmClf.predict(trainData)

    knn_acc=accuracy_score(trainLabel,trainlabel_knn)
    print('knn train accscore:%f'%(knn_acc))

    dt_acc=accuracy_score(trainLabel,trainlabel_dt)
    print('dt train accscore:%f'%(dt_acc))

    rf_acc=accuracy_score(trainLabel,trainlabel_rf)
    print('rf train accscore:%f'%(rf_acc))

    svm_acc=accuracy_score(trainLabel,trainlabel_svm)
    print('svm train accscore:%f'%(svm_acc))

    testLabel_knn=knnClf.predict(testData)
    testLabel_dt=dtClf.predict(testData)
    testLabel_rf=rfClf.predict(testData)
    testLabel_svm=svmClf.predict(testData)

    saveResults(testLabel_knn,os.path.join(data_dir,'output/Result_knn.csv'))
    saveResults(testLabel_dt,os.path.join(data_dir,'output/Result_dt.csv'))
    saveResults(testLabel_rf,os.path.join(data_dir,'output/Result_rf.csv'))
    saveResults(testLabel_svm,os.path.join(data_dir,'output/Result_svm.csv'))
    print('knn dt rf svm process finished ....')
    stop_time2=time.time()
    print('knn classification take:%f' %(stop_time2-start_time))

if __name__=='__main__':
    dRecognition_knn()

最后附knn图：
这里写图片描述
随机森林图：

方法二 cnn待更新：

环境：windows10,64 python报错：
ImportError:no module named tensorflow.python
看来这里涉及到安装Win10下用Anaconda安装TensorFlow

解决方案:TensorFlow目前在Windows下只支持Python 3.5版本。

用Anaconda安装tensorflow, conda命令，这样tensorflow cpu版本就安装好了。
卸载了之前的 python 2.7

测试成功！
cnn code:

#!/usr/bin/python
# -*- coding: utf-8 -*-
"""
Created on Sat Aug 11 11:03:20 2018

@author: Gunther17
"""


import numpy as np
import pandas as pd
from sklearn.model_selection import train_test_split
import os

from keras.callbacks import ReduceLROnPlateau
from keras.layers import Conv2D, Dense, Dropout, Flatten, MaxPool2D
from keras.models import Sequential
from keras.optimizers import RMSprop
from keras.preprocessing.image import ImageDataGenerator
from keras.utils.np_utils import to_categorical  # convert to one-hot-encoding


np.random.seed(2)

# 数据路径
data_dir = 'D:\data Competition\digit-recognizer/'

# Load the data
train = pd.read_csv(os.path.join(data_dir, 'input/train.csv'))
test = pd.read_csv(os.path.join(data_dir, 'input/test.csv'))

X_train = train.values[:, 1:]
Y_train = train.values[:, 0]
test = test.values

# Normalization
X_train = X_train / 255.0
test = test / 255.0

# Reshape image in 3 dimensions (height = 28px, width = 28px , canal = 1)
X_train = X_train.reshape(-1, 28, 28, 1)
test = test.reshape(-1, 28, 28, 1)

# Encode labels to one hot vectors (ex : 2 -> [0,0,1,0,0,0,0,0,0,0])
Y_train = to_categorical(Y_train, num_classes=10)

# Set the random seed
random_seed = 2

# Split the train and the validation set for the fitting
X_train, X_val, Y_train, Y_val = train_test_split(
    X_train, Y_train, test_size=0.1, random_state=random_seed)

# Set the CNN model 
# my CNN architechture is In -> [[Conv2D->relu]*2 -> MaxPool2D -> Dropout]*2 -> Flatten -> Dense -> Dropout -> Out

model = Sequential()

model.add(
    Conv2D(
        filters=32,
        kernel_size=(5, 5),
        padding='Same',
        activation='relu',
        input_shape=(28, 28, 1)))
model.add(
    Conv2D(
        filters=32, kernel_size=(5, 5), padding='Same', activation='relu'))
model.add(MaxPool2D(pool_size=(2, 2)))
model.add(Dropout(0.25))

model.add(
    Conv2D(
        filters=64, kernel_size=(3, 3), padding='Same', activation='relu'))
model.add(
    Conv2D(
        filters=64, kernel_size=(3, 3), padding='Same', activation='relu'))
model.add(MaxPool2D(pool_size=(2, 2), strides=(2, 2)))
model.add(Dropout(0.25))

model.add(Flatten())
model.add(Dense(256, activation="relu"))
model.add(Dropout(0.5))
model.add(Dense(10, activation="softmax"))

# Define the optimizer
optimizer = RMSprop(lr=0.001, rho=0.9, epsilon=1e-08, decay=0.0)

# Compile the model
model.compile(
    optimizer=optimizer, loss="categorical_crossentropy", metrics=["accuracy"])

epochs = 30
batch_size = 86

# Set a learning rate annealer
learning_rate_reduction = ReduceLROnPlateau(
    monitor='val_acc', patience=3, verbose=1, factor=0.5, min_lr=0.00001)

datagen = ImageDataGenerator(
    featurewise_center=False,  # set input mean to 0 over the dataset
    samplewise_center=False,  # set each sample mean to 0
    featurewise_std_normalization=False,  # divide inputs by std of the dataset
    samplewise_std_normalization=False,  # divide each input by its std
    zca_whitening=False,  # apply ZCA whitening
    rotation_range=10,  # randomly rotate images in the range (degrees, 0 to 180)
    zoom_range=0.1,  # Randomly zoom image 
    width_shift_range=0.1,  # randomly shift images horizontally (fraction of total width)
    height_shift_range=0.1,  # randomly shift images vertically (fraction of total height)
    horizontal_flip=False,  # randomly flip images
    vertical_flip=False)  # randomly flip images

datagen.fit(X_train)

history = model.fit_generator(
    datagen.flow(
        X_train, Y_train, batch_size=batch_size),
    epochs=epochs,
    validation_data=(X_val, Y_val),
    verbose=2,
    steps_per_epoch=X_train.shape[0] // batch_size,
    callbacks=[learning_rate_reduction])

# predict results
results = model.predict(test)

# select the indix with the maximum probability
results = np.argmax(results, axis=1)

results = pd.Series(results, name="Label")

submission = pd.concat(
    [pd.Series(
        range(1, 28001), name="ImageId"), results], axis=1)

submission.to_csv(os.path.join(data_dir, 'output/Result_keras_CNN.csv'),index=False)
print('finished')