def make_age_linear(indata, cyto, ax, confounders):
eqn, tmp_cyto = extract_cols(indata, cyto,
['Age']+confounders)
y, X = dmatrices(eqn, tmp_cyto, return_type='dataframe')
model = sm.OLS(y, X).fit()
num = (num for num, col in enumerate(X.columns) if col=='Age').next()
if ax is not None:
plot_fit(model, num, ax=ax)
return model, model.f_pvalue, model.pvalues['Age'], model.params['Age']
def test_plot_oth(self):
#just test that they run
res = self.res
endog = res.model.endog
exog = res.model.exog
plot_fit(res, 0, y_true=None)
plot_partregress(endog, exog, exog_idx=[0, 1])
plot_regress_exog(res, exog_idx=0)
plot_ccpr(res, exog_idx=[0])
plot_ccpr(res, exog_idx=[0, 1])
plt.close('all')
def test_plot_oth(self, close_figures):
#just test that they run
res = self.res
plot_fit(res, 0, y_true=None)
plot_partregress_grid(res, exog_idx=[0,1])
plot_regress_exog(res, exog_idx=0)
plot_ccpr(res, exog_idx=0)
plot_ccpr_grid(res, exog_idx=[0])
fig = plot_ccpr_grid(res, exog_idx=[0,1])
for ax in fig.axes:
add_lowess(ax)
close_or_save(pdf, fig)
def test_plot_oth(self, close_figures):
#just test that they run
res = self.res
plot_fit(res, 0, y_true=None)
plot_partregress_grid(res, exog_idx=[0, 1])
plot_regress_exog(res, exog_idx=0)
plot_ccpr(res, exog_idx=0)
plot_ccpr_grid(res, exog_idx=[0])
fig = plot_ccpr_grid(res, exog_idx=[0, 1])
for ax in fig.axes:
add_lowess(ax)
close_or_save(pdf, fig)
def test_plot_oth(self):
#just test that they run
res = self.res
endog = res.model.endog
exog = res.model.exog
plot_fit(res, 0, y_true=None)
plot_partregress(endog, exog, exog_idx=[0,1])
plot_regress_exog(res, exog_idx=0)
plot_ccpr(res, exog_idx=[0])
plot_ccpr(res, exog_idx=[0,1])
plt.close('all')
def test_plot_oth(self):
#just test that they run
res = self.res
endog = res.model.endog
exog = res.model.exog
plot_fit(res, 0, y_true=None)
plot_partregress_grid(res, exog_idx=[0,1])
plot_regress_exog(res, exog_idx=0)
plot_ccpr(res, exog_idx=0)
plot_ccpr_grid(res, exog_idx=[0])
fig = plot_ccpr_grid(res, exog_idx=[0,1])
for ax in fig.axes:
add_lowess(ax)
plt.close('all')
def test_plot_oth(self):
#just test that they run
res = self.res
endog = res.model.endog
exog = res.model.exog
plot_fit(res, 0, y_true=None)
plot_partregress_grid(res, exog_idx=[0, 1])
plot_regress_exog(res, exog_idx=0)
plot_ccpr(res, exog_idx=0)
plot_ccpr_grid(res, exog_idx=[0])
fig = plot_ccpr_grid(res, exog_idx=[0, 1])
for ax in fig.axes:
add_lowess(ax)
plt.close('all')
def test_plot_fit(self):
res = self.res
fig = plot_fit(res, 0, y_true=None)
x0 = res.model.exog[:, 0]
yf = res.fittedvalues
y = res.model.endog
px1, px2 = fig.axes[0].get_lines()[0].get_data()
np.testing.assert_equal(x0, px1)
np.testing.assert_equal(y, px2)
px1, px2 = fig.axes[0].get_lines()[1].get_data()
np.testing.assert_equal(x0, px1)
np.testing.assert_equal(yf, px2)
plt.close(fig)
def test_plot_fit(self):
res = self.res
fig = plot_fit(res, 0, y_true=None)
x0 = res.model.exog[:, 0]
yf = res.fittedvalues
y = res.model.endog
px1, px2 = fig.axes[0].get_lines()[0].get_data()
np.testing.assert_equal(x0, px1)
np.testing.assert_equal(y, px2)
px1, px2 = fig.axes[0].get_lines()[1].get_data()
np.testing.assert_equal(x0, px1)
np.testing.assert_equal(yf, px2)
close_or_save(pdf, fig)
ax = fig6.add_subplot(2, 1, 1)
#namestr = ' for %s' % self.name if self.name else ''
x1beta = x1 * res.params[1]
x2beta = x2 * res.params[2]
plt.plot(x1, x1beta + res.resid, 'o')
plt.plot(x1, x1beta, '-')
ax.set_title('X_i beta_i plus residuals versus exog (CCPR)') # + namestr)
ax = fig6.add_subplot(2, 1, 2)
plt.plot(x2, x2beta + res.resid, 'o')
plt.plot(x2, x2beta, '-')
#print res.summary()
doplots = 1
if doplots:
fig1 = smrp.plot_fit(res, 0, y_true=None)
smrp.plot_fit(res, 1, y_true=None)
smrp.plot_partregress(y, exog0, exog_idx=[0, 1])
smrp.plot_regress_exog(res, exog_idx=0)
smrp.plot_ccpr(res, exog_idx=[0])
smrp.plot_ccpr(res, exog_idx=[0, 1])
from statsmodels.graphics.tests.test_regressionplots import TestPlot
tp = TestPlot()
tp.test_plot_fit()
fig1 = smrp.plot_partregress(y, exog0, exog_idx=[0, 1])
#add lowess
ax = fig1.axes[0]
y0 = ax.get_lines()[0]._y
x0 = ax.get_lines()[0]._x
####### #Spread Stationarity ###### #Testing for Stationarity on RDA, RDB, Total, Repsol and Centrica #Testing for Stationarity on RDA, RDB concMergered = [RDA['Close'], RDB['Close']] combo = pd.concat(concMergered, axis=1).dropna() combo.columns = ["RDA", "RDB"] model = statsm.ols(formula="RDA~RDB", data=combo) result1 = model.fit() result1.params result1.summary() #plt.plot(result,label='Ordinary least Squares',color='Black') # graph of best fit #statsgraph.plot_fit(result, 1,label='Ordinary least Squares',color='Black') statsgraph.plot_fit(result1, 1) #Testing for Stationarity on Ge, ?, R^2 should be near 1 ( best so far 68% ge on Amazon) # Apple, exxon, Wells no # Microsoft, facebook, WallMart so so #JohnsonJ not great ####### Amazon & Alphabet 68% #Apple on Google 43% # JPM-Wells 67%, Citi no,BOA no, Goldamn 50% # Barcalys - Lloyds 43%-HSBC 62%-Stan 54%-RBS 55% # royal Dutch is stationary hence strong evidence of cointgeration # Credit Suisse - DB 86% #concMergeredCO = [BP['Close'], RDA['Close']] #Another example Centrica and repsol concMergeredCO = [Centrica['Close'], Repsol['Close']]
def test_graphs(self, res):
col_num = self.X.shape[1]
rp.plot_fit(res, exog_idx=col_num - 3)
rp.plot_regress_exog(res, exog_idx=col_num - 1, fig=None)
rp.plot_leverage_resid2(res) # squared
auto.head()
auto_complete = auto.dropna()
auto_complete.shape
# # Linear Regression: mpg ~ weight + price
# ## descriptive statistics
auto['mpg'].describe()
auto[['mpg', 'weight']].describe()
auto[['mpg', 'weight']].mean()
auto[['mpg', 'weight']].std()
sns.pairplot(auto, vars=['mpg', 'weight', 'price'], kind='reg')
model1 = ols('mpg ~ weight + price', data=auto)
fit1 = model1.fit()
fit1.summary()
plot_fit(fit1, 'weight')
fit1.predict()
fit1.resid
# ## Interaction Terms (will include main effects)
ols('mpg ~ weight * price', data=auto).fit().summary()
# ## model with categorical variable & robust standard errors (same as Stata's reg, robust)
# ### [documentation for standard errors](https://www.statsmodels.org/dev/generated/statsmodels.regression.linear_model.RegressionResults.html) (search for HC0_se, HC1_se, etc.)
auto.groupby('foreign').size()
auto.groupby('foreign').mpg.mean()
sns.pairplot(auto, vars=['mpg', 'weight', 'price'], kind='reg', hue='foreign')
fit2 = ols('mpg ~ weight + price + foreign', data=auto).fit(cov_type='HC1')
fit2.summary()
fit2.params
fit2.bse
#namestr = ' for %s' % self.name if self.name else ''
x1beta = x1*res.params[1]
x2beta = x2*res.params[2]
plt.plot(x1, x1beta + res.resid, 'o')
plt.plot(x1, x1beta, '-')
ax.set_title('X_i beta_i plus residuals versus exog (CCPR)')# + namestr)
ax = fig6.add_subplot(2,1,2)
plt.plot(x2, x2beta + res.resid, 'o')
plt.plot(x2, x2beta, '-')
#print res.summary()
doplots = 1
if doplots:
fig1 = smrp.plot_fit(res, 0, y_true=None)
smrp.plot_fit(res, 1, y_true=None)
smrp.plot_partregress_grid(res, exog_idx=[0,1])
smrp.plot_regress_exog(res, exog_idx=0)
smrp.plot_ccpr(res, exog_idx=0)
smrp.plot_ccpr_grid(res, exog_idx=[0,1])
from statsmodels.graphics.tests.test_regressionplots import TestPlot
tp = TestPlot()
tp.test_plot_fit()
fig1 = smrp.plot_partregress_grid(res, exog_idx=[0,1])
#add lowess
ax = fig1.axes[0]
y0 = ax.get_lines()[0]._y
x0 = ax.get_lines()[0]._x
def DailyReturn(Prices):
returns = Prices.shift(1) / Prices - 1
return returns.dropna()
ReturnTotal = DailyReturn(Total['Close'])
ReturnCentrica = DailyReturn(Repsol['Close'])
concMergered = [RDA['Close'], RDB['Close']]
combo = pd.concat(concMergered, axis=1).dropna()
combo.columns = ["RDA", "RDB"]
model = statsm.ols(formula="RDA~RDB", data=combo)
result1 = model.fit()
statsgraph.plot_fit(result1, 1)
concMergeredCO = [Centrica['Close'], Repsol['Close']]
comboCO = pd.concat(concMergeredCO, axis=1).dropna()
comboCO.columns = ["Centrica", "Repsol"]
model = statsm.ols(formula=" Centrica~Repsol", data=comboCO)
resultCO = model.fit()
# Demonstrates clarity for the error term in the model parameters against the residuals
# No drift term
residualsCO = resultCO.resid
concMergeredMV = [
RDA['Close'], RDB['Close'], Total['Close'], Repsol['Close'],
Centrica['Close']
]
