def _var_beta_panel(y, x, beta, xx, rmse, cluster_axis, nw_lags, nobs, df,
nw_overlap):
xx_inv = math.inv(xx)
yv = y.values
if cluster_axis is None:
if nw_lags is None:
return xx_inv * (rmse**2)
else:
resid = yv - np.dot(x.values, beta)
m = (x.values.T * resid).T
xeps = math.newey_west(m, nw_lags, nobs, df, nw_overlap)
return np.dot(xx_inv, np.dot(xeps, xx_inv))
else:
Xb = np.dot(x.values, beta).reshape((len(x.values), 1))
resid = DataFrame(yv[:, None] - Xb, index=y.index, columns=['resid'])
if cluster_axis == 1:
x = x.swaplevel(0, 1).sortlevel(0)
resid = resid.swaplevel(0, 1).sortlevel(0)
m = _group_agg(x.values * resid.values, x.index._bounds,
lambda x: np.sum(x, axis=0))
if nw_lags is None:
nw_lags = 0
xox = 0
for i in range(len(x.index.levels[0])):
xox += math.newey_west(m[i:i + 1], nw_lags, nobs, df, nw_overlap)
return np.dot(xx_inv, np.dot(xox, xx_inv))
def _var_beta_panel(y, x, beta, xx, rmse, cluster_axis, nw_lags, nobs, df, nw_overlap):
from pandas.core.frame import group_agg
xx_inv = math.inv(xx)
yv = y.values
if cluster_axis is None:
if nw_lags is None:
return xx_inv * (rmse ** 2)
else:
resid = yv - np.dot(x.values, beta)
m = (x.values.T * resid).T
xeps = math.newey_west(m, nw_lags, nobs, df, nw_overlap)
return np.dot(xx_inv, np.dot(xeps, xx_inv))
else:
Xb = np.dot(x.values, beta).reshape((len(x.values), 1))
resid = DataFrame(yv[:, None] - Xb, index=y.index, columns=["resid"])
if cluster_axis == 1:
x = x.swaplevel(0, 1).sortlevel(0)
resid = resid.swaplevel(0, 1).sortlevel(0)
m = group_agg(x.values * resid.values, x.index._bounds, lambda x: np.sum(x, axis=0))
if nw_lags is None:
nw_lags = 0
xox = 0
for i in range(len(x.index.levels[0])):
xox += math.newey_west(m[i : i + 1], nw_lags, nobs, df, nw_overlap)
return np.dot(xx_inv, np.dot(xox, xx_inv))
def _cov_beta(self):
cov_resid = self._sigma
x = self._x
inv_cov_x = inv(np.dot(x.T, x))
return np.kron(inv_cov_x, cov_resid)
def _var_beta_raw(self):
"""Returns the raw covariance of beta."""
x = self._x_trans
y = self._y_trans
dates = self._index
nobs = self._nobs
rmse = self._rmse_raw
beta = self._beta_raw
df = self._df_raw
window = self._window
cum_xx = self._cum_xx(self._x)
results = []
for n, i in enumerate(self._valid_indices):
xx = cum_xx[i]
date = dates[i]
if self._is_rolling and i >= window:
xx = xx - cum_xx[i - window]
prior_date = dates[i - window + 1]
else:
prior_date = dates[0]
x_slice = x.truncate(before=prior_date, after=date)
y_slice = y.truncate(before=prior_date, after=date)
xv = x_slice.values
yv = np.asarray(y_slice)
if self._nw_lags is None:
result = math.inv(xx) * (rmse[n]**2)
else:
resid = yv - np.dot(xv, beta[n])
m = (xv.T * resid).T
xeps = math.newey_west(m, self._nw_lags, nobs[n], df[n],
self._nw_overlap)
xx_inv = math.inv(xx)
result = np.dot(xx_inv, np.dot(xeps, xx_inv))
results.append(result)
return np.array(results)
def _var_beta_raw(self):
"""Returns the raw covariance of beta."""
x = self._x_trans
y = self._y_trans
dates = self._index
nobs = self._nobs
rmse = self._rmse_raw
beta = self._beta_raw
df = self._df_raw
window = self._window
cum_xx = self._cum_xx(self._x)
results = []
for n, i in enumerate(self._valid_indices):
xx = cum_xx[i]
date = dates[i]
if self._is_rolling and i >= window:
xx = xx - cum_xx[i - window]
prior_date = dates[i - window + 1]
else:
prior_date = dates[0]
x_slice = x.truncate(before=prior_date, after=date)
y_slice = y.truncate(before=prior_date, after=date)
xv = x_slice.values
yv = np.asarray(y_slice)
if self._nw_lags is None:
result = math.inv(xx) * (rmse[n] ** 2)
else:
resid = yv - np.dot(xv, beta[n])
m = (xv.T * resid).T
xeps = math.newey_west(m, self._nw_lags, nobs[n], df[n],
self._nw_overlap)
xx_inv = math.inv(xx)
result = np.dot(xx_inv, np.dot(xeps, xx_inv))
results.append(result)
return np.array(results)
def _var_beta_raw(self):
"""
Returns the raw covariance of beta.
"""
x = self._x.values
y = self._y.values
xx = np.dot(x.T, x)
if self._nw_lags is None:
return math.inv(xx) * (self._rmse_raw ** 2)
else:
resid = y - np.dot(x, self._beta_raw)
m = (x.T * resid).T
xeps = math.newey_west(m, self._nw_lags, self._nobs, self._df_raw, self._nw_overlap)
xx_inv = math.inv(xx)
return np.dot(xx_inv, np.dot(xeps, xx_inv))
def _var_beta_raw(self):
"""
Returns the raw covariance of beta.
"""
x = self._x.values
y = self._y.values
xx = np.dot(x.T, x)
if self._nw_lags is None:
return math.inv(xx) * (self._rmse_raw**2)
else:
resid = y - np.dot(x, self._beta_raw)
m = (x.T * resid).T
xeps = math.newey_west(m, self._nw_lags, self._nobs, self._df_raw,
self._nw_overlap)
xx_inv = math.inv(xx)
return np.dot(xx_inv, np.dot(xeps, xx_inv))
def _var_beta_panel(y, x, beta, xx, rmse, cluster_axis,
nw_lags, nobs, df, nw_overlap):
from pandas.core.frame import group_agg
from pandas.core.panel import LongPanel
xx_inv = math.inv(xx)
if cluster_axis is None:
if nw_lags is None:
return xx_inv * (rmse ** 2)
else:
resid = y.values.squeeze() - np.dot(x.values, beta)
m = (x.values.T * resid).T
xeps = math.newey_west(m, nw_lags, nobs, df, nw_overlap)
return np.dot(xx_inv, np.dot(xeps, xx_inv))
else:
Xb = np.dot(x.values, beta).reshape((len(x.values), 1))
resid = LongPanel(y.values - Xb, index=y.index,
columns=['resid'])
if cluster_axis == 1:
x = x.swapaxes()
resid = resid.swapaxes()
m = group_agg(x.values * resid.values, x.index._bounds,
lambda x: np.sum(x, axis=0))
if nw_lags is None:
nw_lags = 0
xox = 0
for i in range(len(x.major_axis)):
xox += math.newey_west(m[i : i + 1], nw_lags,
nobs, df, nw_overlap)
return np.dot(xx_inv, np.dot(xox, xx_inv))
def _var_beta_panel(y, x, beta, xx, rmse, cluster_axis, nw_lags, nobs, df,
nw_overlap):
from pandas.core.frame import group_agg
from pandas.core.panel import LongPanel
xx_inv = math.inv(xx)
yv = y.values
if cluster_axis is None:
if nw_lags is None:
return xx_inv * (rmse**2)
else:
resid = yv - np.dot(x.values, beta)
m = (x.values.T * resid).T
xeps = math.newey_west(m, nw_lags, nobs, df, nw_overlap)
return np.dot(xx_inv, np.dot(xeps, xx_inv))
else:
Xb = np.dot(x.values, beta).reshape((len(x.values), 1))
resid = LongPanel(yv[:, None] - Xb, index=y.index, columns=['resid'])
if cluster_axis == 1:
x = x.swapaxes()
resid = resid.swapaxes()
m = group_agg(x.values * resid.values, x.index._bounds,
lambda x: np.sum(x, axis=0))
if nw_lags is None:
nw_lags = 0
xox = 0
for i in range(len(x.major_axis)):
xox += math.newey_west(m[i:i + 1], nw_lags, nobs, df, nw_overlap)
return np.dot(xx_inv, np.dot(xox, xx_inv))
def test_inv_illformed(self):
singular = DataFrame(np.array([[1, 1], [2, 2]]))
rs = pmath.inv(singular)
expected = np.array([[0.1, 0.2], [0.1, 0.2]])
self.assert_(np.allclose(rs, expected))
def test_inv_illformed(self):
singular = DataFrame(np.array([[1, 1], [2, 2]]))
rs = pmath.inv(singular)
expected = np.array([[0.1, 0.2], [0.1, 0.2]])
self.assert_(np.allclose(rs, expected))