notebook.community

Edit and run



In [1]:

    
%pylab inline
from classy import *









    



Populating the interactive namespace from numpy and matplotlib
Version:  0.0.15



In [2]:

    
images=image.load_images('data/digits')









    



[0]: 178 files found
[1]: 182 files found
[2]: 177 files found
[3]: 183 files found
[4]: 181 files found
[5]: 182 files found
[6]: 181 files found
[7]: 179 files found
[8]: 174 files found
[9]: 180 files found



In [3]:

    
data=image.images_to_vectors(images)









    



1797 vectors of length 64
Feature names: 'p0', 'p1', 'p2', 'p3', 'p4'  , ... ,  'p59', 'p60', 'p61', 'p62', 'p63'  (64 features)
Target values given.
Target names: '0', '1', '2', '3', '4', '5', '6', '7', '8', '9'
Mean:  [  0.00000000e+00   4.71341124e+00   8.25158598e+01   1.88401224e+02
   1.88589872e+02   9.17840846e+01   2.15581525e+01   2.04785754e+00
   8.51419032e-02   3.14913745e+01   1.65203116e+02   1.90677240e+02
   1.63504174e+02   1.30023372e+02   2.92036728e+01   1.70005565e+00
   4.22927101e-02   4.11185309e+01   1.57575960e+02   1.11080690e+02
   1.12814691e+02   1.24173066e+02   2.82609905e+01   7.82971619e-01
   1.66944908e-02   3.90011130e+01   1.44642181e+02   1.40286589e+02
   1.58007791e+02   1.20022816e+02   3.66889260e+01   3.33889816e-02
   0.00000000e+00   3.70083472e+01   1.21928770e+02   1.44347245e+02
   1.63978854e+02   1.39081803e+02   4.60222593e+01   0.00000000e+00
   1.37451308e-01   2.49860879e+01   1.09460768e+02   1.14961046e+02
   1.21993879e+02   1.30974958e+02   5.47762938e+01   4.24040067e-01
   1.13522538e-01   1.10378408e+01   1.19312187e+02   1.51698943e+02
   1.49728993e+02   1.39340011e+02   5.90962716e+01   3.23873122e+00
   8.34724541e-03   4.34780189e+00   8.81591541e+01   1.92462994e+02
   1.88009460e+02   1.07495826e+02   3.27490262e+01   5.77072899e+00]
Median:  [   0.    0.   63.  207.  207.   63.    0.    0.    0.    0.  191.  207.
  175.  143.    0.    0.    0.   15.  191.   95.   95.  127.    0.    0.
    0.   15.  175.  159.  191.  111.    0.    0.    0.    0.  127.  159.
  207.  159.   15.    0.    0.    0.   95.  111.  127.  143.   15.    0.
    0.    0.  127.  159.  159.  159.   15.    0.    0.    0.   63.  207.
  223.   95.    0.    0.]
Stddev:  [   0.           14.22717752   75.79479369   67.88517638   68.52232671
   90.3581085    52.88625131   16.47216306    1.44842195   50.7576493
   86.55422116   63.60290344   76.42760101   96.53740043   57.02565325
   13.11647513    0.95321433   56.84679424   90.80932368   92.58050776
   98.50000012   98.78779801   51.79075122    6.90387772    0.50013864
   49.94970201   98.7781811    93.88418435   98.12528728   93.67899482
   58.58726878    0.70690869    0.           55.28329868  100.86027538
   99.97913895   94.64396922   93.64015167   56.19204049    0.
    2.26156209   47.3537032   104.20244968  102.68526906   99.83278433
   90.82739038   68.89733343    4.82912707    3.23145722   27.62316404
   90.02004928   83.47815394   84.68687626   96.17533855   78.31457884
   15.55049191    0.35374992   14.68333191   81.36743295   69.86967274
   78.76953994   94.07347801   65.09626368   29.57005528]



In [4]:

    
data.vectors.shape









    Out[4]:





(1797, 64)



In [5]:

    
data_train,data_test=split(data)









    



Original vector shape:  (1797, 64)
Train vector shape:  (1347, 64)
Test vector shape:  (450, 64)



In [6]:

    
C=NaiveBayes()

timeit(reset=True)
C.fit(data_train.vectors,data_train.targets)
print("Training time: ",timeit())

print("On Training Set:",C.percent_correct(data_train.vectors,data_train.targets))
print("On Test Set:",C.percent_correct(data_test.vectors,data_test.targets))









    



Time Reset
Training time:  0.0044629573822021484 seconds 
On Training Set: 86.4884929473
On Test Set: 85.3333333333

Do the PCA part, reducing the dimensionality



In [7]:

    
F=PCA(5)  # only the top 5



In [8]:

    
data_train_reduced=F.fit_transform_data(data_train)
data_test_reduced=F.transform_data(data_test)



In [9]:

    
train_vectors_reduced=F.fit_transform(data_train.vectors)
test_vectors_reduced=F.transform(data_test.vectors)



In [10]:

    
print("shape train vectors:",data_train.vectors.shape)
print("shape train vectors reduced:",data_train_reduced.vectors.shape)









    



shape train vectors: (1347, 64)
shape train vectors reduced: (1347, 5)



In [11]:

    
timeit(reset=True)
C.fit(data_train_reduced.vectors,data_train_reduced.targets)
print("Training time: ",timeit())

print("On Training Set:",C.percent_correct(data_train_reduced.vectors,data_train_reduced.targets))
print("On Test Set:",C.percent_correct(data_test_reduced.vectors,data_test_reduced.targets))









    



Time Reset
Training time:  0.0032341480255126953 seconds 
On Training Set: 86.1172976986
On Test Set: 84.0



In [12]:

    
F.weights.shape









    Out[12]:





(5, 64)

Visualizing the components



In [13]:

    
F.plot()

you can specify which ones to plot



In [14]:

    
F.plot([2,3,4])

if it is an image, then you can show the components as images



In [15]:

    
F.imshow(shape=(8,8))

Tuning the number of PCs

specify how many PCs to try...



In [16]:

    
PCs=[2,4,6,8,10,20,40]
percent_correct=[]
for n in PCs:
    F=PCA(n) 
    data_train_reduced=F.fit_transform_data(data_train)
    data_test_reduced=F.transform_data(data_test)    
    
    C=NaiveBayes()
    C.fit(data_train_reduced.vectors,data_train_reduced.targets)
    percent_correct.append(C.percent_correct(data_test_reduced.vectors,data_test_reduced.targets))
    
plot(PCs,percent_correct,'-o')
xlabel('Number of PCs')
ylabel('Percent Correct on Test Data')









    Out[16]:





<matplotlib.text.Text at 0x1141e4048>

this does exactly the same thing, but does every number from 1 to 40, skipping every 2 (1,3,5,7,....,39]



In [17]:

    
PCs=arange(1,40,2)
percent_correct=[]
for n in PCs:
    F=PCA(n) 
    data_train_reduced=F.fit_transform_data(data_train)
    data_test_reduced=F.transform_data(data_test)    
    
    C=NaiveBayes()
    C.fit(data_train_reduced.vectors,data_train_reduced.targets)
    percent_correct.append(C.percent_correct(data_test_reduced.vectors,data_test_reduced.targets))
    
plot(PCs,percent_correct,'-o')
xlabel('Number of PCs')
ylabel('Percent Correct on Test Data')









    Out[17]:





<matplotlib.text.Text at 0x114187b38>

Getting rid of some PCs



In [18]:

    
F=PCA(10) 
data_train_reduced=F.fit_transform_data(data_train)
data_test_reduced=F.transform_data(data_test)    

C=NaiveBayes()
C.fit(data_train_reduced.vectors,data_train_reduced.targets)
print(C.percent_correct(data_test_reduced.vectors,data_test_reduced.targets))









    



90.8888888889



In [19]:

    
data_train_reduced_removed=extract_features(data_train_reduced,list(range(2,10)))
data_test_reduced_removed=extract_features(data_test_reduced,list(range(2,10)))

C=NaiveBayes()
C.fit(data_train_reduced_removed.vectors,data_train_reduced_removed.targets)
print(C.percent_correct(data_test_reduced_removed.vectors,data_test_reduced_removed.targets))









    



81.7777777778



In [ ]: