发布 0.7.0¶

import statsmodels.api as sm
from statsmodels.multivariate.pca import PCA

data = sm.datasets.fertility.load_pandas().data

columns = map(str, range(1960, 2012))
data.set_index('Country Name', inplace=True)
dta = data[columns]
dta = dta.dropna()

pca_model = PCA(dta.T, standardize=False, demean=True)
pca_model.plot_scree()

import numpy as np
import statsmodels.api as sm
import pandas as pd

data = sm.datasets.nile.load_pandas().data
data.index = pd.DatetimeIndex(data.year.astype(int).astype(str), freq='AS')

# Setup the state space representation
class LocalLevel(sm.tsa.statespace.MLEModel):
    def __init__(self, endog):
        # Initialize the state space model
        super(LocalLevel, self).__init__(
            endog, k_states=1, initialization='approximate_diffuse')

        # Setup known components of state space representation matrices
        self.ssm['design', :] = 1.
        self.ssm['transition', :] = 1.
        self.ssm['selection', :] = 1.

    # Describe how parameters enter the model
    def update(self, params, transformed=True):
        params = super(LocalLevel, self).update(params, transformed)
        self.ssm['obs_cov', 0, 0] = params[0]
        self.ssm['state_cov', 0, 0] = params[1]

    def transform_params(self, params):
        return params**2  # force variance parameters to be positive

    # Specify start parameters and parameter names
    @property
    def start_params(self):
        return [np.std(self.endog)]*2

    @property
    def param_names(self):
        return ['sigma2.measurement', 'sigma2.level']

# Fit the model with maximum likelihood estimation
mod = LocalLevel(data['volume'])
res = mod.fit()
print res.summary()

import statsmodels.api as sm
import pandas as pd

data = sm.datasets.macrodata.load_pandas().data
data.index = pd.DatetimeIndex(start='1959-01-01', end='2009-09-01',
                              freq='QS')
endog = data['realcons']
exog = data['m1']

mod = sm.tsa.SARIMAX(endog, exog=exog, order=(1,1,1),
                     trend='t', seasonal_order=(0,0,1,4))
res = mod.fit()
print res.summary()