In [1]:

    

import numpy as np
import pandas as pd
import models.imports.features

import seaborn as sns
import matplotlib.pyplot as plt

sns.set()


    

In [2]:

    

## load in features df.

trainsf_ = r'../data/text/bitcoin/train_set.csv'
features_df = models.imports.features.import_file(trainsf_)


    

In [3]:

    

features_df.info()


    

    




<class 'pandas.core.frame.DataFrame'>
Int64Index: 1754 entries, 0 to 1753
Data columns (total 6 columns):
date     1754 non-null object
chibs    1754 non-null float64
hm       1754 non-null float64
is       1754 non-null float64
lr       1754 non-null float64
price    1754 non-null float64
dtypes: float64(5), object(1)
memory usage: 95.9+ KB

In [4]:

    

## split into x, y

features_set_x, features_set_y = models.imports.features.create_dataset(features_df)


    

In [5]:

    

## x,


    

In [6]:

    

features_set_x.describe()


    

    
Out[6]:







  

    

      

      
chibs
      
hm
      
is
      
lr
    
  
  

    

      
count
      
1754.000000
      
1754.000000
      
1754.000000
      
1754.000000
    
    

      
mean
      
-56396.219603
      
-68234.384726
      
-77587.501197
      
-56291.299614
    
    

      
std
      
51940.257460
      
60791.875443
      
69651.745734
      
50620.714335
    
    

      
min
      
-550446.937040
      
-608469.455649
      
-690089.181171
      
-509062.460009
    
    

      
25%
      
-74211.183197
      
-90299.695851
      
-103232.827310
      
-74332.227083
    
    

      
50%
      
-50000.412654
      
-60584.970105
      
-68693.517044
      
-50135.630238
    
    

      
75%
      
-24743.757304
      
-31645.638258
      
-35440.577409
      
-25263.345517
    
    

      
max
      
-664.394640
      
-727.634661
      
-606.808984
      
-532.879345
    
  

In [7]:

    

features_set_x.plot(alpha=.5, title='estimates')


    

    
Out[7]:





<matplotlib.axes._subplots.AxesSubplot at 0x1d240661c88>






    

In [8]:

    

features_set_x.plot.hist(bins=100, alpha=.5, title='estimate histograms')


    

    
Out[8]:





<matplotlib.axes._subplots.AxesSubplot at 0x1d2407ae278>






    

In [9]:

    

## y,


    

In [10]:

    

features_set_y.describe()


    

    
Out[10]:







  

    

      

      
price
    
  
  

    

      
count
      
1754.000000
    
    

      
mean
      
967.288005
    
    

      
std
      
2299.649478
    
    

      
min
      
4.270000
    
    

      
25%
      
109.132500
    
    

      
50%
      
375.050000
    
    

      
75%
      
654.950000
    
    

      
max
      
19343.040000
    
  

In [11]:

    

## train output,


    

In [12]:

    

plt.subplot(1,4,1)
plt.title('norm')
plt.plot(features_set_y)

plt.subplot(1,4,2)
plt.title('log')
plt.plot(np.log(features_set_y))

plt.subplot(1,4,3)
plt.title('scaled')
plt.plot(features_set_y / np.max(features_set_y))

plt.subplot(1,4,4)
plt.title('log + scaled')
transformed_y = np.log(features_set_y)
transformed_y /= np.max(transformed_y)
plt.plot(transformed_y)


plt.tight_layout()
plt.show()


    

In [13]:

    

features_set_y.plot.hist(bins=100, title='price histogram')


    

    
Out[13]:





<matplotlib.axes._subplots.AxesSubplot at 0x1d240afb2b0>