In [14]:

    

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

import seaborn as sns
import matplotlib.pyplot as plt

sns.set()


    

In [15]:

    

## load in features df.

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


    

In [16]:

    

features_df.info()


    

    




<class 'pandas.core.frame.DataFrame'>
Int64Index: 438 entries, 0 to 437
Data columns (total 6 columns):
date     438 non-null object
chibs    438 non-null float64
hm       438 non-null float64
is       438 non-null float64
lr       438 non-null float64
price    438 non-null float64
dtypes: float64(5), object(1)
memory usage: 24.0+ KB

In [17]:

    

## split into x, y

test_set_x, test_set_y = models.imports.features.create_dataset(features_df)


    

In [18]:

    

## x,


    

In [19]:

    

test_set_x.describe()


    

    
Out[19]:







  

    

      

      
chibs
      
hm
      
is
      
lr
    
  
  

    

      
count
      
438.000000
      
438.000000
      
438.000000
      
438.000000
    
    

      
mean
      
-55153.808507
      
-66529.265613
      
-75349.367168
      
-55028.698495
    
    

      
std
      
42663.375130
      
49726.140700
      
57495.218033
      
41716.623213
    
    

      
min
      
-305415.252186
      
-352072.199048
      
-413326.708742
      
-292502.817959
    
    

      
25%
      
-74187.056052
      
-89739.540644
      
-102376.014141
      
-73206.271557
    
    

      
50%
      
-50830.446764
      
-63141.625600
      
-70437.452087
      
-52113.932467
    
    

      
75%
      
-25949.610707
      
-33241.088109
      
-37184.234472
      
-26441.975005
    
    

      
max
      
-838.793301
      
-1095.388574
      
-1061.871666
      
-911.180683
    
  

In [20]:

    

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


    

    
Out[20]:





<matplotlib.axes._subplots.AxesSubplot at 0x2013b4ca320>






    

In [21]:

    

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


    

    
Out[21]:





<matplotlib.axes._subplots.AxesSubplot at 0x2013b51fcf8>






    

In [22]:

    

## y,


    

In [23]:

    

test_set_y.describe()


    

    
Out[23]:







  

    

      

      
price
    
  
  

    

      
count
      
438.000000
    
    

      
mean
      
756.124110
    
    

      
std
      
1309.982576
    
    

      
min
      
4.220000
    
    

      
25%
      
122.087500
    
    

      
50%
      
379.710000
    
    

      
75%
      
674.965000
    
    

      
max
      
10895.010000
    
  

In [24]:

    

## train output,


    

In [25]:

    

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

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

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

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


plt.tight_layout()
plt.show()


    

In [26]:

    

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


    

    
Out[26]:





<matplotlib.axes._subplots.AxesSubplot at 0x2013b61f160>