In [9]:

    

import pandas as pd
import matplotlib.pyplot as plt

from matplotlib import axes
from matplotlib import rc

%matplotlib inline
font = {'family' : 'monospace',
        'weight' : 'bold',
        'size'   : 20}

rc('font', **font)
import


    

In [2]:

    

ionosphere = pd.read_csv("results/ionosphere_results.csv", index_col="Unnamed: 0")
ionosphere['dataset'] = 'ionosphere'
ionosphere['dim_reduction'] = 33 - ionosphere['encoding_dim']
ionosphere['accuracy_reduction'] = ionosphere['neural_net_acc'] - ionosphere['auto_classifier_acc']
ionosphere.head()


    

    
Out[2]:







  

    

      

      
auto_classifier_acc
      
encoding_dim
      
neural_net_acc
      
dataset
      
dim_reduction
      
accuracy_reduction
    
  
  

    

      
0
      
0.942857
      
33
      
0.857143
      
ionosphere
      
0
      
-0.085714
    
    

      
1
      
0.914286
      
32
      
0.857143
      
ionosphere
      
1
      
-0.057143
    
    

      
2
      
0.857143
      
31
      
0.857143
      
ionosphere
      
2
      
0.000000
    
    

      
3
      
0.857143
      
30
      
0.857143
      
ionosphere
      
3
      
0.000000
    
    

      
4
      
0.942857
      
29
      
0.857143
      
ionosphere
      
4
      
-0.085714
    
  

In [3]:

    

robo = pd.read_csv("results/robo_results.csv", index_col="Unnamed: 0")
robo['dataset'] = 'robo'
robo['dim_reduction'] = 24 - robo['encoding_dim']
robo['accuracy_reduction'] = robo['neural_net_acc'] - robo['auto_classifier_acc']
robo.head()


    

    
Out[3]:







  

    

      

      
auto_classifier_acc
      
encoding_dim
      
neural_net_acc
      
dataset
      
dim_reduction
      
accuracy_reduction
    
  
  

    

      
0
      
0.939560
      
24
      
0.945055
      
robo
      
0
      
0.005495
    
    

      
1
      
0.939560
      
23
      
0.945055
      
robo
      
1
      
0.005495
    
    

      
2
      
0.935897
      
22
      
0.945055
      
robo
      
2
      
0.009158
    
    

      
3
      
0.959707
      
21
      
0.945055
      
robo
      
3
      
-0.014652
    
    

      
4
      
0.928571
      
20
      
0.945055
      
robo
      
4
      
0.016484
    
  

In [4]:

    

semeion = pd.read_csv("results/semeion_results.csv", index_col="Unnamed: 0")
semeion['dataset'] = 'semeion'
semeion['dim_reduction'] = 40 - semeion['encoding_dim']
semeion['accuracy_reduction'] = semeion['neural_net_acc'] - semeion['auto_classifier_acc']
semeion.head()


    

    
Out[4]:







  

    

      

      
auto_classifier_acc
      
encoding_dim
      
neural_net_acc
      
dataset
      
dim_reduction
      
accuracy_reduction
    
  
  

    

      
0
      
0.918239
      
40
      
0.91195
      
semeion
      
0
      
-6.289308e-03
    
    

      
1
      
0.899371
      
39
      
0.91195
      
semeion
      
1
      
1.257861e-02
    
    

      
2
      
0.937107
      
38
      
0.91195
      
semeion
      
2
      
-2.515724e-02
    
    

      
3
      
0.911950
      
37
      
0.91195
      
semeion
      
3
      
0.000000e+00
    
    

      
4
      
0.911950
      
36
      
0.91195
      
semeion
      
4
      
-5.623080e-09
    
  

In [5]:

    

results = pd.concat([ionosphere, robo, semeion], axis=0, ignore_index=True)
results.to_csv("results/combined_results")


    

In [11]:

    

results.iloc[results.groupby('dataset').accuracy_reduction.idxmin()]


    

    
Out[11]:







  

    

      

      
auto_classifier_acc
      
encoding_dim
      
neural_net_acc
      
dataset
      
dim_reduction
      
accuracy_reduction
    
  
  

    

      
0
      
0.942857
      
33
      
0.857143
      
ionosphere
      
0
      
-0.085714
    
    

      
36
      
0.959707
      
21
      
0.945055
      
robo
      
3
      
-0.014652
    
    

      
59
      
0.937107
      
38
      
0.911950
      
semeion
      
2
      
-0.025157
    
  

In [12]:

    

results[results['accuracy_reduction'] < 0]


    

    
Out[12]:







  

    

      

      
auto_classifier_acc
      
encoding_dim
      
neural_net_acc
      
dataset
      
dim_reduction
      
accuracy_reduction
    
  
  

    

      
0
      
0.942857
      
33
      
0.857143
      
ionosphere
      
0
      
-8.571429e-02
    
    

      
1
      
0.914286
      
32
      
0.857143
      
ionosphere
      
1
      
-5.714286e-02
    
    

      
4
      
0.942857
      
29
      
0.857143
      
ionosphere
      
4
      
-8.571429e-02
    
    

      
5
      
0.942857
      
28
      
0.857143
      
ionosphere
      
5
      
-8.571429e-02
    
    

      
6
      
0.914286
      
27
      
0.857143
      
ionosphere
      
6
      
-5.714286e-02
    
    

      
8
      
0.942857
      
25
      
0.857143
      
ionosphere
      
8
      
-8.571429e-02
    
    

      
9
      
0.914286
      
24
      
0.857143
      
ionosphere
      
9
      
-5.714286e-02
    
    

      
10
      
0.914286
      
23
      
0.857143
      
ionosphere
      
10
      
-5.714286e-02
    
    

      
11
      
0.942857
      
22
      
0.857143
      
ionosphere
      
11
      
-8.571429e-02
    
    

      
12
      
0.942857
      
21
      
0.857143
      
ionosphere
      
12
      
-8.571429e-02
    
    

      
14
      
0.914286
      
19
      
0.857143
      
ionosphere
      
14
      
-5.714286e-02
    
    

      
15
      
0.942857
      
18
      
0.857143
      
ionosphere
      
15
      
-8.571429e-02
    
    

      
17
      
0.885714
      
16
      
0.857143
      
ionosphere
      
17
      
-2.857143e-02
    
    

      
18
      
0.885714
      
15
      
0.857143
      
ionosphere
      
18
      
-2.857143e-02
    
    

      
19
      
0.885714
      
14
      
0.857143
      
ionosphere
      
19
      
-2.857143e-02
    
    

      
20
      
0.885714
      
13
      
0.857143
      
ionosphere
      
20
      
-2.857143e-02
    
    

      
21
      
0.914286
      
12
      
0.857143
      
ionosphere
      
21
      
-5.714286e-02
    
    

      
22
      
0.885714
      
11
      
0.857143
      
ionosphere
      
22
      
-2.857143e-02
    
    

      
23
      
0.942857
      
10
      
0.857143
      
ionosphere
      
23
      
-8.571429e-02
    
    

      
24
      
0.942857
      
9
      
0.857143
      
ionosphere
      
24
      
-8.571429e-02
    
    

      
25
      
0.885714
      
8
      
0.857143
      
ionosphere
      
25
      
-2.857143e-02
    
    

      
27
      
0.914286
      
6
      
0.857143
      
ionosphere
      
27
      
-5.714286e-02
    
    

      
28
      
0.885714
      
5
      
0.857143
      
ionosphere
      
28
      
-2.857143e-02
    
    

      
29
      
0.914286
      
4
      
0.857143
      
ionosphere
      
29
      
-5.714286e-02
    
    

      
36
      
0.959707
      
21
      
0.945055
      
robo
      
3
      
-1.465201e-02
    
    

      
57
      
0.918239
      
40
      
0.911950
      
semeion
      
0
      
-6.289308e-03
    
    

      
59
      
0.937107
      
38
      
0.911950
      
semeion
      
2
      
-2.515724e-02
    
    

      
61
      
0.911950
      
36
      
0.911950
      
semeion
      
4
      
-5.623080e-09
    
    

      
62
      
0.924528
      
35
      
0.911950
      
semeion
      
5
      
-1.257862e-02
    
    

      
63
      
0.937107
      
34
      
0.911950
      
semeion
      
6
      
-2.515723e-02
    
    

      
64
      
0.930818
      
33
      
0.911950
      
semeion
      
7
      
-1.886792e-02
    
    

      
65
      
0.918239
      
32
      
0.911950
      
semeion
      
8
      
-6.289308e-03
    
    

      
69
      
0.930818
      
28
      
0.911950
      
semeion
      
12
      
-1.886793e-02
    
    

      
70
      
0.930818
      
27
      
0.911950
      
semeion
      
13
      
-1.886792e-02
    
    

      
71
      
0.924528
      
26
      
0.911950
      
semeion
      
14
      
-1.257862e-02
    
    

      
73
      
0.924528
      
24
      
0.911950
      
semeion
      
16
      
-1.257862e-02
    
    

      
74
      
0.930818
      
23
      
0.911950
      
semeion
      
17
      
-1.886793e-02
    
    

      
76
      
0.911950
      
21
      
0.911950
      
semeion
      
19
      
-5.623080e-09
    
    

      
77
      
0.918239
      
20
      
0.911950
      
semeion
      
20
      
-6.289308e-03
    
    

      
79
      
0.918239
      
18
      
0.911950
      
semeion
      
22
      
-6.289308e-03
    
    

      
83
      
0.918239
      
14
      
0.911950
      
semeion
      
26
      
-6.289309e-03
    
    

      
84
      
0.918239
      
13
      
0.911950
      
semeion
      
27
      
-6.289313e-03
    
    

      
85
      
0.930818
      
12
      
0.911950
      
semeion
      
28
      
-1.886792e-02
    
  

In [ ]:

    

fig = plt.figure(figsize=(20,10))
plt.plot(model.history.history['val_acc'])
plt.plot(model.history.history['acc'])
plt.axhline(y=nn_score, c="red")
plt.text(0, nn_score, "test: " + str(round(nn_score, 4)), fontdict=font)
plt.title('model accuracy for neural net with 2 hidden layers')
plt.ylabel('accuracy')
plt.xlabel('epochs')
plt.legend(['valid', 'train'], loc='lower right')
plt.show()


    

In [14]:

    

fig = plt.figure(figsize=(20,10))

plt.bar(results['accuracy_reduction'])


    

    




---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-14-63cf67195831> in <module>()
      1 fig = plt.figure(figsize=(20,10))
      2 
----> 3 plt.bar(results['accuracy_reduction'])

C:\ProgramData\Anaconda3\lib\site-packages\matplotlib\pyplot.py in bar(*args, **kwargs)
   2625                       mplDeprecation)
   2626     try:
-> 2627         ret = ax.bar(*args, **kwargs)
   2628     finally:
   2629         ax._hold = washold

C:\ProgramData\Anaconda3\lib\site-packages\matplotlib\__init__.py in inner(ax, *args, **kwargs)
   1708                     warnings.warn(msg % (label_namer, func.__name__),
   1709                                   RuntimeWarning, stacklevel=2)
-> 1710             return func(ax, *args, **kwargs)
   1711         pre_doc = inner.__doc__
   1712         if pre_doc is None:

C:\ProgramData\Anaconda3\lib\site-packages\matplotlib\axes\_axes.py in bar(self, *args, **kwargs)
   2025                 break
   2026         else:
-> 2027             raise exps[0]
   2028         # if we matched the second-case, then the user passed in
   2029         # left=val as a kwarg which we want to deprecate

C:\ProgramData\Anaconda3\lib\site-packages\matplotlib\axes\_axes.py in bar(self, *args, **kwargs)
   2017         for matcher in matchers:
   2018             try:
-> 2019                 dp, x, height, width, y, kwargs = matcher(*args, **kwargs)
   2020             except TypeError as e:
   2021                 # This can only come from a no-match as there is

TypeError: <lambda>() missing 1 required positional argument: 'height'