Kaggle (digit recognizer) PCA+SVM

下载minst数据集

kaggle competitions download -c digit-recognizer

包含两个文件：

• train.csv
• test.csv

import pandas as pd
import matplotlib.pyplot as plt
import numpy as np

train = pd.read_csv('../../../datasets/digit-recognizer/train.csv')
print(train.shape)
test = pd.read_csv('../../../datasets/digit-recognizer/test.csv')
print(test.shape)

(42000, 785)
(28000, 784)

print(train.describe)

<bound method NDFrame.describe of        label  pixel0  pixel1  pixel2  pixel3  pixel4  pixel5  pixel6  pixel7  \
0          1       0       0       0       0       0       0       0       0   
1          0       0       0       0       0       0       0       0       0   
2          1       0       0       0       0       0       0       0       0   
3          4       0       0       0       0       0       0       0       0   
4          0       0       0       0       0       0       0       0       0   
...      ...     ...     ...     ...     ...     ...     ...     ...     ...   
41995      0       0       0       0       0       0       0       0       0   
41996      1       0       0       0       0       0       0       0       0   
41997      7       0       0       0       0       0       0       0       0   
41998      6       0       0       0       0       0       0       0       0   
41999      9       0       0       0       0       0       0       0       0   

       pixel8  ...  pixel774  pixel775  pixel776  pixel777  pixel778  \
0           0  ...         0         0         0         0         0   
1           0  ...         0         0         0         0         0   
2           0  ...         0         0         0         0         0   
3           0  ...         0         0         0         0         0   
4           0  ...         0         0         0         0         0   
...       ...  ...       ...       ...       ...       ...       ...   
41995       0  ...         0         0         0         0         0   
41996       0  ...         0         0         0         0         0   
41997       0  ...         0         0         0         0         0   
41998       0  ...         0         0         0         0         0   
41999       0  ...         0         0         0         0         0   

       pixel779  pixel780  pixel781  pixel782  pixel783  
0             0         0         0         0         0  
1             0         0         0         0         0  
2             0         0         0         0         0  
3             0         0         0         0         0  
4             0         0         0         0         0  
...         ...       ...       ...       ...       ...  
41995         0         0         0         0         0  
41996         0         0         0         0         0  
41997         0         0         0         0         0  
41998         0         0         0         0         0  
41999         0         0         0         0         0  

[42000 rows x 785 columns]>

y_train = train['label']
train_label = y_train.values.astype(int)
X_train = train.drop('label', axis=1)
X_train = X_train.values.astype(int)
X_test = test.values.astype(int)

X_example = X_train[0].reshape(28, -1)
%matplotlib inline
plt.imshow(X_example, cmap='gray')

<matplotlib.image.AxesImage at 0x26e59836e48>

from sklearn.pipeline import Pipeline
from sklearn.preprocessing import StandardScaler
from sklearn.svm import SVC
from sklearn.decomposition import PCA

PCA+SVM

pca = PCA(whiten=True, n_components=0.95)

pca.fit(X_train)
train_data = pca.transform(X_train)

# clf_svm = SVC(kernel='rbf', gamma=5, C=0.001)
clf_svm = SVC()
clf_svm.fit(train_data, train_label)

test_data = pca.transform(X_test)
test_predict = clf_svm.predict(test_data)

pca_svm_submission = pd.DataFrame(
    {'ImageId':range(1, 28001), 'Label':test_predict}
)

pca_svm_submission.to_csv('pca_svm_submission.csv', index=False)
print(pca_svm_submission)

       ImageId  Label
0            1      2
1            2      0
2            3      9
3            4      7
4            5      3
...        ...    ...
27995    27996      9
27996    27997      7
27997    27998      3
27998    27999      9
27999    28000      2

[28000 rows x 2 columns]