#data_concat.plot.scatter('LSTAT','MEDV')
lstat = data[['LSTAT']]
lm = LinearRegression()
lm.fit(lstat, target)
lm.score(lstat, target)
lstat = sm.add_constant(lstat)
lm = sm.OLS(target, lstat).fit()
lm.summary()
lm.params
lstat_plot = np.linspace(0, 40)
plt.scatter(lstat.LSTAT, target)
plt.plot(lstat_plot, lstat_plot * lm.params[1] + lm.params[0], color='red')
lm_fitted_y = lm.fittedvalues
lm_residuals = lm.resid
lm_norm_residuals = lm.get_influence().resid_studentized_internal
lm_norm_resid_abs_sqrt = np.sqrt(np.abs(lm_norm_residuals))
lm_abs_resid = np.abs(lm_residuals)
lm_leverage = lm.get_influence().hat_matrix_diag
lm_cooks = lm.get_influence().cooks_distance[0]
sns.residplot(lm_fitted_y, lm.resid, lowess=True)
sm.qqplot(lm_norm_residuals, fit=True)
plt.scatter(lm_fitted_y, lm_norm_resid_abs_sqrt)
sns.regplot(lm_fitted_y, lm_norm_resid_abs_sqrt, lowess=True)
sns.regplot(lm_leverage, lm_norm_residuals, lowess=True)
