In [1]:

    

## imports,
import numpy as np
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

import models.imports.features

sns.set()


    

In [2]:

    

## load in features df.

trainsf_ = r'../data/text/bitcoin/train_set.csv'
train_set = models.imports.features.import_file(trainsf_)
train_set_x, train_set_y = models.imports.features.scale_and_transform_into_datasets(train_set)


    

In [3]:

    

## train features,

train_set_x.plot(alpha=.5)


    

    
Out[3]:





<matplotlib.axes._subplots.AxesSubplot at 0x2d17b970630>






    

In [4]:

    

## output features,

train_set_y.plot()


    

    
Out[4]:





<matplotlib.axes._subplots.AxesSubplot at 0x2d17bae8710>






    

In [5]:

    

from sklearn import linear_model


    

In [6]:

    

model = linear_model.LinearRegression()
model.fit(train_set_x, train_set_y)

prediction = model.predict(train_set_x)
mse = models.imports.features.mse(train_set_y, prediction)
print('mse:', mse)


    

    




mse: price    0.006382
dtype: float64

In [7]:

    

plt.plot(model.predict(train_set_x), 'ro', markersize=1)
plt.plot(train_set_y, 'b', alpha=.3)
plt.show()


    

In [8]:

    

## test,

testsf_ = r'../data/text/bitcoin/test_set.csv'
test_set = models.imports.features.import_file(testsf_)
test_set_x, test_set_y = models.imports.features.scale_and_transform_into_datasets(test_set)


    

In [9]:

    

## mean squared error,
test_prediction = model.predict(test_set_x)
mse_test = models.imports.features.mse(test_set_y, test_prediction)
print('mse:', mse_test)


    

    




mse: price    0.01199
dtype: float64

In [10]:

    

plt.plot(model.predict(test_set_x), 'ro', markersize=1)
plt.plot(test_set_y, 'b', alpha=.3)
plt.show()


    

In [ ]: