import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import statsmodels.tsa.stattools as ts
data=pd.read_csv('C:/Users/HXWD/Desktop/数据/rb.csv',encoding='gbk')
data.columns=['date','open','high','low','close','amt','opi']
data.head()
np.log(data['close']).head()
x = np.array(np.log(data['close']))
result = ts.adfuller(x, 1,regresults=True) # maxlag is now set to 1
print(result)

#结果：

(-1.0159755159305488, 0.74739309585919544, {'1%': -3.4336805486772994, '10%': -2.5675532292926859, '5%': -2.8630112431183181}, 

<statsmodels.tsa.stattools.ResultsStore object at 0x0A9C3CD0>)

#以上计算的价格的对数的单位根检验，检验结果不显著，存在单位根。但是计算每天对数收益率的时候，这个是不存在单位根的。

adfuller(x, maxlag=None, regression='c', autolag='AIC', store=False, regresults=False)
    Augmented Dickey-Fuller unit root test
    
    The Augmented Dickey-Fuller test can be used to test for a unit root in a
    univariate process in the presence of serial correlation.
    
    Parameters
    ----------
    x : array_like, 1d
        data series
    maxlag : int
        Maximum lag which is included in test, default 12*(nobs/100)^{1/4}
    regression : str {'c','ct','ctt','nc'}
        Constant and trend order to include in regression
        * 'c' : constant only (default)
        * 'ct' : constant and trend
        * 'ctt' : constant, and linear and quadratic trend
        * 'nc' : no constant, no trend
    autolag : {'AIC', 'BIC', 't-stat', None}
        * if None, then maxlag lags are used
        * if 'AIC' (default) or 'BIC', then the number of lags is chosen
          to minimize the corresponding information criterium
        * 't-stat' based choice of maxlag.  Starts with maxlag and drops a
          lag until the t-statistic on the last lag length is significant at
          the 95 % level.
    store : bool
        If True, then a result instance is returned additionally to
        the adf statistic (default is False)
    regresults : bool
        If True, the full regression results are returned (default is False)
    
    Returns
    -------
    adf : float
        Test statistic
    pvalue : float
        MacKinnon's approximate p-value based on MacKinnon (1994)
    usedlag : int
        Number of lags used.
    nobs : int
        Number of observations used for the ADF regression and calculation of
        the critical values.
    critical values : dict
        Critical values for the test statistic at the 1 %, 5 %, and 10 %
        levels. Based on MacKinnon (2010)
    icbest : float
        The maximized information criterion if autolag is not None.
    regresults : RegressionResults instance
        The
    resstore : (optional) instance of ResultStore
        an instance of a dummy class with results attached as attributes
    
    Notes
    -----
    The null hypothesis of the Augmented Dickey-Fuller is that there is a unit
    root, with the alternative that there is no unit root. If the pvalue is
    above a critical size, then we cannot reject that there is a unit root.
    
    The p-values are obtained through regression surface approximation from
    MacKinnon 1994, but using the updated 2010 tables.
    If the p-value is close to significant, then the critical values should be
    used to judge whether to accept or reject the null.
    
    The autolag option and maxlag for it are described in Greene.
    
    Examples
    --------
    see example script
    
    References
    ----------
    Greene
    Hamilton
    
    
    P-Values (regression surface approximation)
    MacKinnon, J.G. 1994.  "Approximate asymptotic distribution functions for
    unit-root and cointegration tests.  `Journal of Business and Economic
    Statistics` 12, 167-76.
    
    Critical values
    MacKinnon, J.G. 2010. "Critical Values for Cointegration Tests."  Queen's
    University, Dept of Economics, Working Papers.  Available at
    http://ideas.repec.org/p/qed/wpaper/1227.html