03 Python Implementation.html

<p>
  In python, we have a very power package for mathematical models, which is named 'statsmodels'.
</p>
<div class="section-example-container">

<pre class="python">import statsmodels.formula.api as sm
model = sm.ols(formula = 'amzn~spy',data = df).fit()
print model.summary()
</pre>
</div>

<img class="img-responsive" src="https://cdn.quantconnect.com/tutorials/i/Tutorial09-plot2.png" alt="plot2" />
<p>
  We built a simple linear regression model above by using the ols() function in statsmodels. The 'model' instance has lots of properties. The most commonly used one is parameters, or slope and intercept. We can access to them by:
</p>
<div class="section-example-container">

<pre class="python">print 'pamameters: ',model.params
[out]: pamameters:  Intercept    0.000012
                   spy          0.492112
print 'residual: ', model.resid.tail()
[out]: residual:  Date
2016-12-22   -0.003360
2016-12-23   -0.008219
2016-12-28    0.000477
2016-12-29   -0.006303
2016-12-30   -0.021211
print 'fitted values: ',model.predict()
[out]: fitted values:  [-0.00070299 -0.00218348  0.00068734  0.00046907 -0.00277819  0.00103882]
</pre>
</div>

<p>
  Now let's have a look at our fitted line:
</p>
<div class="section-example-container">

<pre class="python">plt.figure(figsize = (15,10))
plt.scatter(df.spy,df.amzn)
plt.xlabel('spx_return')
plt.ylabel('amzn_return')
plt.plot(df.spy,model.predict(),color = 'red')
plt.show()
</pre>
</div>

<img class="img-responsive" src="https://cdn.quantconnect.com/tutorials/i/Tutorial09-head.png" alt="head" />
<p>
  The red line is the fitted linear regression straight line. As we can see there are lot of statistical results in the summary table. Now let's talk about some important statistical parameters.
</p>
<img class="img-responsive" src="https://cdn.quantconnect.com/tutorials/i/Tutorial09-regression-table.png" alt="regression table" />