x\c@ sdZddlmZddlmZddlmZddlmZddl Z ddl Z ddl j jZddljZddlmZddlmZmZmZdd lmZmZmZmZmZmZm Z m!Z!m"Z"dd l#m$Z$m%Z%m&Z&m'Z'm(Z(m)Z)dd l*m+Z+m,Z,ddl-j.j/Z0dd l1m2Z2dd l3m4Z4e5e2Z2dZ6de+e,fdYZ7de7fdYZ8de4fdYZ9de7fdYZ:de:fdYZ;de;fdYZ<de9e<fdYZ=de;fdYZ>de9e>fd YZ?d!Z@d"ZAd#e:fd$YZBeCd%ZDd&ZEd'ZFd(ZGd)ZHd*ZId+ZJeKd,ZLe8jeL_d-ZMe>jeM_d.ZNeBjeN_dS(/sc Provide a generic structure to support window functions, similar to how we have a Groupby object. i(tdivision(t defaultdict(t timedelta(tdedentN(tfunction(tAppendert Substitutiontcache_readonly( tensure_float64tis_booltis_float_dtypet is_integertis_integer_dtypet is_list_liket is_scalartis_timedelta64_dtypetneeds_i8_conversion(t ABCDataFramet ABCDateOffsettABCDatetimeIndextABCPeriodIndext ABCSeriestABCTimedeltaIndex(t PandasObjecttSelectionMixin(t _shared_docs(t GroupByMixins Returns ------- Series or DataFrame Return type is determined by the caller. See Also -------- Series.%(name)s : Series %(name)s. DataFrame.%(name)s : DataFrame %(name)s. t_WindowcB sWeZdddddddgZeZd!d!ed!dd!d!dZed Z ed Z ed Z ed Z d Z dZdZd!dZdZdZd!dZedZdZdZd!dZd!edZd!d!dZdZdZdZeZede dRR t _selection(R+tkeyR>tsubset((s1lib/python2.7/site-packages/pandas/core/window.pyt_gotitemxs   $cC sY||jkrtj||S||jkr6||Stdt|j|fdS(Ns%r object has no attribute %r(t_internal_names_settobjectt__getattribute__R&tAttributeErrorttypet__name__(R+tattr((s1lib/python2.7/site-packages/pandas/core/window.pyt __getattr__s cC s |jjS(N(R&t_dir_additions(R+((s1lib/python2.7/site-packages/pandas/core/window.pyRSscC s|jS(N(R(R+tother((s1lib/python2.7/site-packages/pandas/core/window.pyt _get_windowscC s |jjS(N(t __class__RP(R+((s1lib/python2.7/site-packages/pandas/core/window.pyt _window_typesc C sqg|jD]?}t||ddk r djd|dt||^q }djd|jddj|S( s@ Provide a nice str repr of our rolling object. s{k}={v}tktvs{klass} [{attrs}]tklasstattrst,N(t _attributestgetattrR'tformatRWtjoin(R+RXR[((s1lib/python2.7/site-packages/pandas/core/window.pyt __unicode__s  ?cC s"d}tdjd|dS(Ns1https://github.com/pandas-dev/pandas/issues/11704sSee issue #11704 {url}turl(tNotImplementedErrorR_(R+Rb((s1lib/python2.7/site-packages/pandas/core/window.pyt__iter__scC s8|jr.|dkr!|j}n||jfS||fS(sw Return index as ndarrays. Returns ------- tuple of (index, index_as_ndarray) N(R3R'R1tasi8(R+R:((s1lib/python2.7/site-packages/pandas/core/window.pyt _get_indexs    cC s |dkr't|jd|j}nt|jrEt|}nt|jrct|}nxt|jrtdj d|j d|jnByt|}Wn/t t fk rt dj |jnX|r|j }tj|tj| {0}(R'R^R9R RhRR RRcR_RWR8t TypeErrorR=tnptNaNtisinf(R+RCtkill_inf((s1lib/python2.7/site-packages/pandas/core/window.pyt _prep_valuess&     cC s|d kr|j}n|j}t|tjr|d k rt|jjrddl m }||j ddjj |j }qn|jdkrddl m}|||d|jSt||d|d |jS|S( s' Wrap a single result. i(t to_timedeltatunittnsi(tSeriestnameR:R<N(R'R9R:t isinstanceRjtndarrayRRCRhtpandasRotraveltreshapetshapeR>RrRsROR<(R+tresulttblockR&R:RoRr((s1lib/python2.7/site-packages/pandas/core/window.pyt _wrap_results    *cC sddlm}m}ddlm}g}xXt||D]G\}} |j|d| d|}|jdkrv|S|j|q<W|j j } |j d k rY|j j|j rY|j j} |j||j d|jd| |jd k rY||j} | | krV|jj } | j| j| g} | jt| } qVqYnt|sr|jd S||d djd | d tS(s Wrap the results. Parameters ---------- results : list of ndarrays blocks : list of blocks obj : conformed data (may be resampled) i(Rrtconcat(t ensure_indexR{R&iR:Rstfloat64R R<R=N(RvRrR}tpandas.core.indexR~tzipR|R>tappendR9R<R!R'R1tequalsR:RsRGR&t get_indexerttolistttaketsortedtlentastypeR?RA(R+tresultsR%R&RrR}R~tfinalRzR{R<Rst selectiontindexer((s1lib/python2.7/site-packages/pandas/core/window.pyt _wrap_resultss*  % %    cC s|j|jdkr%tdnt|t}|dkrt|ttfrq|j| d|j}qt dg|j}t |d||jR8t_offsettTrueRtRRt slice_shifttsliceR'RjR=ttuple(R+RzRtoffsett lead_indexer((s1lib/python2.7/site-packages/pandas/core/window.pyt_center_window%s cO sJ|j|||\}}|dkrF|j|dtd|d|S|S(NtrawtargsR,(t _aggregateR'tapplyRA(R+targRR,Rzthow((s1lib/python2.7/site-packages/pandas/core/window.pyt aggregate7s s* Calculate %(name)s sum of given DataFrame or Series. Parameters ---------- *args, **kwargs For compatibility with other %(name)s methods. Has no effect on the computed value. Returns ------- Series or DataFrame Same type as the input, with the same index, containing the %(name)s sum. See Also -------- Series.sum : Reducing sum for Series. DataFrame.sum : Reducing sum for DataFrame. Examples -------- >>> s = pd.Series([1, 2, 3, 4, 5]) >>> s 0 1 1 2 2 3 3 4 4 5 dtype: int64 >>> s.rolling(3).sum() 0 NaN 1 NaN 2 6.0 3 9.0 4 12.0 dtype: float64 >>> s.expanding(3).sum() 0 NaN 1 NaN 2 6.0 3 10.0 4 15.0 dtype: float64 >>> s.rolling(3, center=True).sum() 0 NaN 1 6.0 2 9.0 3 12.0 4 NaN dtype: float64 For DataFrame, each %(name)s sum is computed column-wise. >>> df = pd.DataFrame({"A": s, "B": s ** 2}) >>> df A B 0 1 1 1 2 4 2 3 9 3 4 16 4 5 25 >>> df.rolling(3).sum() A B 0 NaN NaN 1 NaN NaN 2 6.0 14.0 3 9.0 29.0 4 12.0 50.0 tsums Calculate the %(name)s mean of the values. Parameters ---------- *args Under Review. **kwargs Under Review. Returns ------- Series or DataFrame Returned object type is determined by the caller of the %(name)s calculation. See Also -------- Series.%(name)s : Calling object with Series data. DataFrame.%(name)s : Calling object with DataFrames. Series.mean : Equivalent method for Series. DataFrame.mean : Equivalent method for DataFrame. Examples -------- The below examples will show rolling mean calculations with window sizes of two and three, respectively. >>> s = pd.Series([1, 2, 3, 4]) >>> s.rolling(2).mean() 0 NaN 1 1.5 2 2.5 3 3.5 dtype: float64 >>> s.rolling(3).mean() 0 NaN 1 NaN 2 2.0 3 3.0 dtype: float64 tmeanN(!RPt __module__R]tsett exclusionsR'RAR-tpropertyR/R0R1R3R*R;RDRJRRRSRURWRaRdRfRRnR|RRRtaggRR(((s1lib/python2.7/site-packages/pandas/core/window.pyR-s<          .  I *R.c B seZdZdZdZedZedZedZ e dede dd d d d d e e d dZ e Ze dde e ddZe dde e ddZRS(sh Provides rolling window calculations. .. versionadded:: 0.18.0 Parameters ---------- window : int, or offset Size of the moving window. This is the number of observations used for calculating the statistic. Each window will be a fixed size. If its an offset then this will be the time period of each window. Each window will be a variable sized based on the observations included in the time-period. This is only valid for datetimelike indexes. This is new in 0.19.0 min_periods : int, default None Minimum number of observations in window required to have a value (otherwise result is NA). For a window that is specified by an offset, `min_periods` will default to 1. Otherwise, `min_periods` will default to the size of the window. center : bool, default False Set the labels at the center of the window. win_type : str, default None Provide a window type. If ``None``, all points are evenly weighted. See the notes below for further information. on : str, optional For a DataFrame, column on which to calculate the rolling window, rather than the index axis : int or str, default 0 closed : str, default None Make the interval closed on the 'right', 'left', 'both' or 'neither' endpoints. For offset-based windows, it defaults to 'right'. For fixed windows, defaults to 'both'. Remaining cases not implemented for fixed windows. .. versionadded:: 0.20.0 Returns ------- a Window or Rolling sub-classed for the particular operation See Also -------- expanding : Provides expanding transformations. ewm : Provides exponential weighted functions. Notes ----- By default, the result is set to the right edge of the window. This can be changed to the center of the window by setting ``center=True``. To learn more about the offsets & frequency strings, please see `this link `__. The recognized win_types are: * ``boxcar`` * ``triang`` * ``blackman`` * ``hamming`` * ``bartlett`` * ``parzen`` * ``bohman`` * ``blackmanharris`` * ``nuttall`` * ``barthann`` * ``kaiser`` (needs beta) * ``gaussian`` (needs std) * ``general_gaussian`` (needs power, width) * ``slepian`` (needs width). If ``win_type=None`` all points are evenly weighted. To learn more about different window types see `scipy.signal window functions `__. Examples -------- >>> df = pd.DataFrame({'B': [0, 1, 2, np.nan, 4]}) >>> df B 0 0.0 1 1.0 2 2.0 3 NaN 4 4.0 Rolling sum with a window length of 2, using the 'triang' window type. >>> df.rolling(2, win_type='triang').sum() B 0 NaN 1 1.0 2 2.5 3 NaN 4 NaN Rolling sum with a window length of 2, min_periods defaults to the window length. >>> df.rolling(2).sum() B 0 NaN 1 1.0 2 3.0 3 NaN 4 NaN Same as above, but explicitly set the min_periods >>> df.rolling(2, min_periods=1).sum() B 0 0.0 1 1.0 2 3.0 3 2.0 4 4.0 A ragged (meaning not-a-regular frequency), time-indexed DataFrame >>> df = pd.DataFrame({'B': [0, 1, 2, np.nan, 4]}, ... index = [pd.Timestamp('20130101 09:00:00'), ... pd.Timestamp('20130101 09:00:02'), ... pd.Timestamp('20130101 09:00:03'), ... pd.Timestamp('20130101 09:00:05'), ... pd.Timestamp('20130101 09:00:06')]) >>> df B 2013-01-01 09:00:00 0.0 2013-01-01 09:00:02 1.0 2013-01-01 09:00:03 2.0 2013-01-01 09:00:05 NaN 2013-01-01 09:00:06 4.0 Contrasting to an integer rolling window, this will roll a variable length window corresponding to the time period. The default for min_periods is 1. >>> df.rolling('2s').sum() B 2013-01-01 09:00:00 0.0 2013-01-01 09:00:02 1.0 2013-01-01 09:00:03 3.0 2013-01-01 09:00:05 NaN 2013-01-01 09:00:06 4.0 cC stt|j|j}t|tttjfr:nt |r|dkrat dnyddl j }Wnt k rt dnXt|jtjst dj|jnt||jddkrt dj|jqnt dj|dS(Niswindow must be > 0 is.Please install scipy to generate window weightsInvalid win_type {0}sInvalid window {0}(tsuperR.R*RRttlistRRjRuR R8t scipy.signaltsignalt ImportErrorRtcompatt string_typesR_R^R'(R+Rtsig((s1lib/python2.7/site-packages/pandas/core/window.pyR*Ns     c s|j}t|tttjfr=tj|jt St |rddl j }fd}d||j |}|j||tjt SdS(ss Provide validation for our window type, return the window we have already been validated. iNc sbidgd6dgd6ddgd6dgd6}||kr^t|g||||S|S( Ntbetatkaisertstdtgaussiantpowertwidthtgeneral_gaussiantslepian(R(RR,targ_map(t _pop_args(s1lib/python2.7/site-packages/pandas/core/window.pyt_validate_win_typeqs      cS sZd|}g}xC|D];}||kr<t||n|j|j|qW|S(Ns%s window requires %%s(R8Rtpop(Rt arg_namesR,tmsgtall_argstn((s1lib/python2.7/site-packages/pandas/core/window.pyR|s   (RURtRRRjRutcomtasarray_tuplesafeRtfloatR RRRt get_windowRA(R+R,RRRR((Rs1lib/python2.7/site-packages/pandas/core/window.pyt _prep_windowds   c  sEj|jj\}}}g}x|D]}yj|j}Wn*tk r|j|jjq:nX|jdkr|j|jq:nt } t j t j g| fd} t j | j|} r!j| } n|j| q:Wj|||S(s Applies a moving window of type ``window_type`` on the data. Parameters ---------- mean : bool, default True If True computes weighted mean, else weighted sum Returns ------- y : same type as input argument ic sLtjt}tjr9tj|fn||dS(Ntavg(t _use_windowRRt libwindowt roll_windowRjt concatenate(RRR,tminp(tadditional_nansRRR+R(s1lib/python2.7/site-packages/pandas/core/window.pytfs'(RRRDRnRCRiRR=tsizeRRjtarrayRktapply_along_axisR RR( R+RR,R%R&R:RtbRCRRRz((RRRR+Rs1lib/python2.7/site-packages/pandas/core/window.pyt _apply_windows*   se See Also -------- pandas.DataFrame.rolling.aggregate pandas.DataFrame.aggregate s Examples -------- >>> df = pd.DataFrame(np.random.randn(10, 3), columns=['A', 'B', 'C']) >>> df A B C 0 -2.385977 -0.102758 0.438822 1 -1.004295 0.905829 -0.954544 2 0.735167 -0.165272 -1.619346 3 -0.702657 -1.340923 -0.706334 4 -0.246845 0.211596 -0.901819 5 2.463718 3.157577 -1.380906 6 -1.142255 2.340594 -0.039875 7 1.396598 -1.647453 1.677227 8 -0.543425 1.761277 -0.220481 9 -0.640505 0.289374 -1.550670 >>> df.rolling(3, win_type='boxcar').agg('mean') A B C 0 NaN NaN NaN 1 NaN NaN NaN 2 -0.885035 0.212600 -0.711689 3 -0.323928 -0.200122 -1.093408 4 -0.071445 -0.431533 -1.075833 5 0.504739 0.676083 -0.996353 6 0.358206 1.903256 -0.774200 7 0.906020 1.283573 0.085482 8 -0.096361 0.818139 0.472290 9 0.070889 0.134399 -0.031308 tsee_alsotexamplest versionaddedtRZsSeries/DataFrameR RcO s:|j|||\}}|dkr6||}n|S(N(RR'(R+RRR,RzR((s1lib/python2.7/site-packages/pandas/core/window.pyRs RsRRcO s&tjd|||jdt|S(NRR(tnvtvalidate_window_funcRRA(R+RR,((s1lib/python2.7/site-packages/pandas/core/window.pyRsRcO s&tjd|||jdt|S(NR(RRRR(R+RR,((s1lib/python2.7/site-packages/pandas/core/window.pyRs(RPRt__doc__R*RRRRt_agg_see_also_doct_agg_examples_docRRRRRRR(((s1lib/python2.7/site-packages/pandas/core/window.pyR.s&  % 0      t _GroupByMixincB sneZdZdZejdZejdddddZejdddddZ ddddZ RS( s) Provide the groupby facilities. cO s|jdd}|jdd}|dkrC||j}}n||_t|j_t|jj_tt|j |||dS(Ntparenttgroupby( RR'R&t_groupbyRtmutatedtgrouperRRR-(R+R&RR,RR((s1lib/python2.7/site-packages/pandas/core/window.pyR-s   tcounttcorrRTtpairwisetcovc s%|fd}jj|S(s Dispatch to apply; we are stripping all of the _apply kwargs and performing the original function call on the grouped object. c sJj|}t|tjr7t|||S|j||S(N(RERtRRR^R(txRsR(R,R+(s1lib/python2.7/site-packages/pandas/core/window.pyRs(RR(R+tfuncRsRRt check_minpR,R((R,R+s1lib/python2.7/site-packages/pandas/core/window.pyt_applysN( RPRRR-Rt _dispatchRR'RRR(((s1lib/python2.7/site-packages/pandas/core/window.pyRs t_RollingcB s,eZedZdddddZRS(cC stS(N(tRolling(R+((s1lib/python2.7/site-packages/pandas/core/window.pyR/%sc s|d krj}nd kr3jnd krHtnj\}}} jd| \} g} x|D]} j| j} | jdkr| j | j qnt t j r.ttd d kr tdjdnd d fdn|rwt|} tjtjg| fd}nfd}tjdd ;| jd krtj|j| }n || }Wd QX|rj|}n| j |qWj| ||S( s Rolling statistical measure using supplied function. Designed to be used with passed-in Cython array-based functions. Parameters ---------- func : str/callable to apply name : str, optional name of this function window : int/array, default to _get_window() center : bool, default to self.center check_minp : function, default to _use_window Returns ------- y : type of input R:is3we do not support this function in libwindow.{func}Rc s4||}t|}||||S(N(R(RRRR"R(tcfuncRtindexiR,(s1lib/python2.7/site-packages/pandas/core/window.pyRVs c s.tj|fdjdjS(NRR"(RjRRR"(R(RRR+R(s1lib/python2.7/site-packages/pandas/core/window.pytcalcbsc s|djdjS(NRR"(RR"(R(RR+R(s1lib/python2.7/site-packages/pandas/core/window.pyRhstalltignoreiN(R'RRURRDRfRnRCRRR=RtRRR^RR8R_RRjRRkterrstateR>RR RR(R+RRsRRRR,R%R&R:RRRCRRRz((RRRRRR,R+Rs1lib/python2.7/site-packages/pandas/core/window.pyR)sB        !N(RPRRR/R'R(((s1lib/python2.7/site-packages/pandas/core/window.pyR#s t_Rolling_and_ExpandingcB s[eZeded>> s = pd.Series([2, 3, np.nan, 10]) >>> s.rolling(2).count() 0 1.0 1 2.0 2 1.0 3 1.0 dtype: float64 >>> s.rolling(3).count() 0 1.0 1 2.0 2 2.0 3 2.0 dtype: float64 >>> s.rolling(4).count() 0 1.0 1 2.0 2 2.0 3 3.0 dtype: float64 Rc C s|j\}}}|jd||j}|jsOt|t|n|}g}xc|D][}|jjt}|j |d|ddd|jd|j j }|j |qbW|j |||S(NR:RRiRR"(RDRfRURtminRtnotnaRtintR/R"RRR(R+R%R&R:RRRRz((s1lib/python2.7/site-packages/pandas/core/window.pyRs $  s The %(name)s function's apply function. Parameters ---------- func : function Must produce a single value from an ndarray input if ``raw=True`` or a Series if ``raw=False``. raw : bool, default None * ``False`` : passes each row or column as a Series to the function. * ``True`` or ``None`` : the passed function will receive ndarray objects instead. If you are just applying a NumPy reduction function this will achieve much better performance. The `raw` parameter is required and will show a FutureWarning if not passed. In the future `raw` will default to False. .. versionadded:: 0.23.0 *args, **kwargs Arguments and keyword arguments to be passed into func. Returns ------- Series or DataFrame Return type is determined by the caller. See Also -------- Series.%(name)s : Series %(name)s. DataFrame.%(name)s : DataFrame %(name)s. Rc  sddlmjdd}j}t|jj\}dkr}tj dt ddt nfd}j |dd d t d S( Ni(Rrt_levelsCurrently, 'apply' passes the values as ndarrays to the applied function. In the future, this will change to passing it as Series objects. You need to specify 'raw=True' to keep the current behaviour, and you can pass 'raw=False' to silence this warningt stacklevelic  sXt||}s0|djj}ntj|||| S(NR:(RR&R:Rt roll_generic(RRRR"R(RrRRRR,RRR+(s1lib/python2.7/site-packages/pandas/core/window.pyRs  RR,RR(RvRrRR'RURRRftwarningstwarnt FutureWarningRRRA( R+RRRR,RRR:R((RrRRRR,RRR+s1lib/python2.7/site-packages/pandas/core/window.pyRs    $cO s&tjd|||jdd|S(NRtroll_sum(RRR(R+RR,((s1lib/python2.7/site-packages/pandas/core/window.pyRss Calculate the %(name)s maximum. Parameters ---------- *args, **kwargs Arguments and keyword arguments to be passed into func. tmaxcO s&tjd|||jdd|S(NRtroll_max(RRR(R+RR,((s1lib/python2.7/site-packages/pandas/core/window.pyRss Calculate the %(name)s minimum. Parameters ---------- **kwargs Under Review. Returns ------- Series or DataFrame Returned object type is determined by the caller of the %(name)s calculation. See Also -------- Series.%(name)s : Calling object with a Series. DataFrame.%(name)s : Calling object with a DataFrame. Series.min : Similar method for Series. DataFrame.min : Similar method for DataFrame. Examples -------- Performing a rolling minimum with a window size of 3. >>> s = pd.Series([4, 3, 5, 2, 6]) >>> s.rolling(3).min() 0 NaN 1 NaN 2 3.0 3 2.0 4 2.0 dtype: float64 RcO s&tjd|||jdd|S(NRtroll_min(RRR(R+RR,((s1lib/python2.7/site-packages/pandas/core/window.pyR*scO s&tjd|||jdd|S(NRt roll_mean(RRR(R+RR,((s1lib/python2.7/site-packages/pandas/core/window.pyR.ss Calculate the %(name)s median. Parameters ---------- **kwargs For compatibility with other %(name)s methods. Has no effect on the computed median. Returns ------- Series or DataFrame Returned type is the same as the original object. See Also -------- Series.%(name)s : Calling object with Series data. DataFrame.%(name)s : Calling object with DataFrames. Series.median : Equivalent method for Series. DataFrame.median : Equivalent method for DataFrame. Examples -------- Compute the rolling median of a series with a window size of 3. >>> s = pd.Series([0, 1, 2, 3, 4]) >>> s.rolling(3).median() 0 NaN 1 NaN 2 1.0 3 2.0 4 3.0 dtype: float64 tmediancK s|jdd|S(Nt roll_median_cR(R(R+R,((s1lib/python2.7/site-packages/pandas/core/window.pyRUss Calculate %(name)s standard deviation. Normalized by N-1 by default. This can be changed using the `ddof` argument. Parameters ---------- ddof : int, default 1 Delta Degrees of Freedom. The divisor used in calculations is ``N - ddof``, where ``N`` represents the number of elements. *args, **kwargs For NumPy compatibility. No additional arguments are used. Returns ------- Series or DataFrame Returns the same object type as the caller of the %(name)s calculation. See Also -------- Series.%(name)s : Calling object with Series data. DataFrame.%(name)s : Calling object with DataFrames. Series.std : Equivalent method for Series. DataFrame.std : Equivalent method for DataFrame. numpy.std : Equivalent method for Numpy array. Notes ----- The default `ddof` of 1 used in Series.std is different than the default `ddof` of 0 in numpy.std. A minimum of one period is required for the rolling calculation. Examples -------- >>> s = pd.Series([5, 5, 6, 7, 5, 5, 5]) >>> s.rolling(3).std() 0 NaN 1 NaN 2 0.577350 3 1.000000 4 1.000000 5 1.154701 6 0.000000 dtype: float64 >>> s.expanding(3).std() 0 NaN 1 NaN 2 0.577350 3 0.957427 4 0.894427 5 0.836660 6 0.786796 dtype: float64 Ric sntjd||jj\}fd}j|ddtdd|S(NRc s=tdj}ttj||jS(Ni(t_require_min_periodsRt_zsqrtRtroll_varR"(RRR,R(tddofRR+R(s1lib/python2.7/site-packages/pandas/core/window.pyRsRiR(RRRURfRR(R+RRR,R:R((RRR+Rs1lib/python2.7/site-packages/pandas/core/window.pyRs  s Calculate unbiased %(name)s variance. Normalized by N-1 by default. This can be changed using the `ddof` argument. Parameters ---------- ddof : int, default 1 Delta Degrees of Freedom. The divisor used in calculations is ``N - ddof``, where ``N`` represents the number of elements. *args, **kwargs For NumPy compatibility. No additional arguments are used. Returns ------- Series or DataFrame Returns the same object type as the caller of the %(name)s calculation. See Also -------- Series.%(name)s : Calling object with Series data. DataFrame.%(name)s : Calling object with DataFrames. Series.var : Equivalent method for Series. DataFrame.var : Equivalent method for DataFrame. numpy.var : Equivalent method for Numpy array. Notes ----- The default `ddof` of 1 used in :meth:`Series.var` is different than the default `ddof` of 0 in :func:`numpy.var`. A minimum of 1 period is required for the rolling calculation. Examples -------- >>> s = pd.Series([5, 5, 6, 7, 5, 5, 5]) >>> s.rolling(3).var() 0 NaN 1 NaN 2 0.333333 3 1.000000 4 1.000000 5 1.333333 6 0.000000 dtype: float64 >>> s.expanding(3).var() 0 NaN 1 NaN 2 0.333333 3 0.916667 4 0.800000 5 0.700000 6 0.619048 dtype: float64 tvarcO s8tjd|||jdddtdd||S(NRRRiR(RRRR(R+RRR,((s1lib/python2.7/site-packages/pandas/core/window.pyRss Unbiased %(name)s skewness. Parameters ---------- **kwargs Keyword arguments to be passed into func. tskewcK s|jdddtd|S(Nt roll_skewRRi(RR(R+R,((s1lib/python2.7/site-packages/pandas/core/window.pyRss Calculate unbiased %(name)s kurtosis. This function uses Fisher's definition of kurtosis without bias. Parameters ---------- **kwargs Under Review. Returns ------- Series or DataFrame Returned object type is determined by the caller of the %(name)s calculation See Also -------- Series.%(name)s : Calling object with Series data. DataFrame.%(name)s : Calling object with DataFrames. Series.kurt : Equivalent method for Series. DataFrame.kurt : Equivalent method for DataFrame. scipy.stats.skew : Third moment of a probability density. scipy.stats.kurtosis : Reference SciPy method. Notes ----- A minimum of 4 periods is required for the %(name)s calculation. tkurtcK s|jdddtd|S(Nt roll_kurtRRi(RR(R+R,((s1lib/python2.7/site-packages/pandas/core/window.pyR ss Calculate the %(name)s quantile. Parameters ---------- quantile : float Quantile to compute. 0 <= quantile <= 1. interpolation : {'linear', 'lower', 'higher', 'midpoint', 'nearest'} .. versionadded:: 0.23.0 This optional parameter specifies the interpolation method to use, when the desired quantile lies between two data points `i` and `j`: * linear: `i + (j - i) * fraction`, where `fraction` is the fractional part of the index surrounded by `i` and `j`. * lower: `i`. * higher: `j`. * nearest: `i` or `j` whichever is nearest. * midpoint: (`i` + `j`) / 2. **kwargs: For compatibility with other %(name)s methods. Has no effect on the result. Returns ------- Series or DataFrame Returned object type is determined by the caller of the %(name)s calculation. See Also -------- pandas.Series.quantile : Computes value at the given quantile over all data in Series. pandas.DataFrame.quantile : Computes values at the given quantile over requested axis in DataFrame. Examples -------- >>> s = pd.Series([1, 2, 3, 4]) >>> s.rolling(2).quantile(.4, interpolation='lower') 0 NaN 1 1.0 2 2.0 3 3.0 dtype: float64 >>> s.rolling(2).quantile(.4, interpolation='midpoint') 0 NaN 1 1.5 2 2.5 3 3.5 dtype: float64 tquantiletlinearc sRjj\}fd}j|dd|S(Nc stj}dkr:tj||jSdkrbtj||jStj||jSdS(Ng?g(RRRRR"Rt roll_quantile(RRR,R(Rt interpolationR R+R(s1lib/python2.7/site-packages/pandas/core/window.pyRHs     R (RURfR(R+R R R,R:R((RR R R+Rs1lib/python2.7/site-packages/pandas/core/window.pyR Ds   s Calculate the %(name)s sample covariance. Parameters ---------- other : Series, DataFrame, or ndarray, optional If not supplied then will default to self and produce pairwise output. pairwise : bool, default None If False then only matching columns between self and other will be used and the output will be a DataFrame. If True then all pairwise combinations will be calculated and the output will be a MultiIndexed DataFrame in the case of DataFrame inputs. In the case of missing elements, only complete pairwise observations will be used. ddof : int, default 1 Delta Degrees of Freedom. The divisor used in calculations is ``N - ddof``, where ``N`` represents the number of elements. **kwargs Keyword arguments to be passed into func. Rc s|dkr0j}|dkr'tn|}nj|}jrTjnj|fd}tj|j|dt|S(Nc s|jd}|jd}fd}||jddjj}||}||||||||S(NRc s%|jjdjjS(NR(trollingRRR(R(R,R+R(s1lib/python2.7/site-packages/pandas/core/window.pytsRR(RR RR(tXtYRRtbias_adj(RR,R+R(s1lib/python2.7/site-packages/pandas/core/window.pyt_get_cov{sR( R'R9RRER3R(RUt_flex_binary_momenttbool(R+RTRRR,R((RR,R+Rs1lib/python2.7/site-packages/pandas/core/window.pyRns     s Calculate %(name)s correlation. Parameters ---------- other : Series, DataFrame, or ndarray, optional If not supplied then will default to self. pairwise : bool, default None Calculate pairwise combinations of columns within a DataFrame. If `other` is not specified, defaults to `True`, otherwise defaults to `False`. Not relevant for :class:`~pandas.Series`. **kwargs Unused. Returns ------- Series or DataFrame Returned object type is determined by the caller of the %(name)s calculation. See Also -------- Series.%(name)s : Calling object with Series data. DataFrame.%(name)s : Calling object with DataFrames. Series.corr : Equivalent method for Series. DataFrame.corr : Equivalent method for DataFrame. %(name)s.cov : Similar method to calculate covariance. numpy.corrcoef : NumPy Pearson's correlation calculation. Notes ----- This function uses Pearson's definition of correlation (https://en.wikipedia.org/wiki/Pearson_correlation_coefficient). When `other` is not specified, the output will be self correlation (e.g. all 1's), except for :class:`~pandas.DataFrame` inputs with `pairwise` set to `True`. Function will return ``NaN`` for correlations of equal valued sequences; this is the result of a 0/0 division error. When `pairwise` is set to `False`, only matching columns between `self` and `other` will be used. When `pairwise` is set to `True`, the output will be a MultiIndex DataFrame with the original index on the first level, and the `other` DataFrame columns on the second level. In the case of missing elements, only complete pairwise observations will be used. Examples -------- The below example shows a rolling calculation with a window size of four matching the equivalent function call using :meth:`numpy.corrcoef`. >>> v1 = [3, 3, 3, 5, 8] >>> v2 = [3, 4, 4, 4, 8] >>> fmt = "{0:.6f}" # limit the printed precision to 6 digits >>> # numpy returns a 2X2 array, the correlation coefficient >>> # is the number at entry [0][1] >>> print(fmt.format(np.corrcoef(v1[:-1], v2[:-1])[0][1])) 0.333333 >>> print(fmt.format(np.corrcoef(v1[1:], v2[1:])[0][1])) 0.916949 >>> s1 = pd.Series(v1) >>> s2 = pd.Series(v2) >>> s1.rolling(4).corr(s2) 0 NaN 1 NaN 2 NaN 3 0.333333 4 0.916949 dtype: float64 The below example shows a similar rolling calculation on a DataFrame using the pairwise option. >>> matrix = np.array([[51., 35.], [49., 30.], [47., 32.], [46., 31.], [50., 36.]]) >>> print(np.corrcoef(matrix[:-1,0], matrix[:-1,1]).round(7)) [[1. 0.6263001] [0.6263001 1. ]] >>> print(np.corrcoef(matrix[1:,0], matrix[1:,1]).round(7)) [[1. 0.5553681] [0.5553681 1. ]] >>> df = pd.DataFrame(matrix, columns=['X','Y']) >>> df X Y 0 51.0 35.0 1 49.0 30.0 2 47.0 32.0 3 46.0 31.0 4 50.0 36.0 >>> df.rolling(4).corr(pairwise=True) X Y 0 X NaN NaN Y NaN NaN 1 X NaN NaN Y NaN NaN 2 X NaN NaN Y NaN NaN 3 X 1.000000 0.626300 Y 0.626300 1.000000 4 X 1.000000 0.555368 Y 0.555368 1.000000 Rc s|dkr0j}|dkr'tn|}nj|}j|fd}tj|j|dt|S(Nc sr|jddjdj}|jddjdj}|j||j|jS(NRRR(R RRRR(taR(R,R+R(s1lib/python2.7/site-packages/pandas/core/window.pyt _get_corrs   R(R'R9RRERURR(R+RTRR,R((R,R+Rs1lib/python2.7/site-packages/pandas/core/window.pyRs  N((RPRRRRR'RRRRRRRRRRR RR(((s1lib/python2.7/site-packages/pandas/core/window.pyRzsH$      !   !  8  8     4  ( k Rc B s,eZedZedZdZdZdZedZ edZ e de de d d d d d d e e ddZeZe dde e ddZe dde e dd1d2idZe dde e ddZe dde ee e ddZe dde e ddZe dde e ddZe dde e ddZe dde e d d!d"Ze dde e d#d!d$Ze dde ee e d%d&Zed'Ze ee dde e d(d)Ze dde e d*d+d,Ze dde ee e d-d1d1d!d.Ze dde e d/d1d1d0Z RS(3cC st|jtttfS(N(RtR1RRR(R+((s1lib/python2.7/site-packages/pandas/core/window.pyR0 s cC s|jdkr|jjSt|jtrd|j|jjkrdddlm}||j|jSt dj |jdS(Ni(tIndexsDinvalid on specified as {0}, must be a column (if DataFrame) or None( R!R'R&R:RtRR<RvRR8R_(R+R((s1lib/python2.7/site-packages/pandas/core/window.pyR1s  cC stt|j|jjs(|jrt|jtj t t fr|j |j }|jrttdn|j|_|j|_d|_|jdkrd|_qn<t|jstdn|jdkrtdn|j r|jdk rtdndS(NsCcenter is not implemented for datetimelike and offset based windowsR2iswindow must be an integeriswindow must be non-negativesAclosed only implemented for datetimelike and offset based windows(RRR*R&temptyR0RtRRRRRt_validate_monotonict_validate_freqRRcR(tnanosRRR'R R8R"(R+R2((s1lib/python2.7/site-packages/pandas/core/window.pyR*"s&       cC s7|jjs3|jpd}tdj|ndS(s+ Validate on is_monotonic. R:s{0} must be monotonicN(R1t is_monotonicR!R8R_(R+t formatted((s1lib/python2.7/site-packages/pandas/core/window.pyREs  cC sWddlm}y||jSWn/ttfk rRtdj|jnXdS(s5 Validate & return window frequency. i(t to_offsets=passed window {0} is not compatible with a datetimelike indexN(tpandas.tseries.frequenciesRRRiR8R_(R+R((s1lib/python2.7/site-packages/pandas/core/window.pyRNs  sV See Also -------- pandas.Series.rolling pandas.DataFrame.rolling s Examples -------- >>> df = pd.DataFrame(np.random.randn(10, 3), columns=['A', 'B', 'C']) >>> df A B C 0 -2.385977 -0.102758 0.438822 1 -1.004295 0.905829 -0.954544 2 0.735167 -0.165272 -1.619346 3 -0.702657 -1.340923 -0.706334 4 -0.246845 0.211596 -0.901819 5 2.463718 3.157577 -1.380906 6 -1.142255 2.340594 -0.039875 7 1.396598 -1.647453 1.677227 8 -0.543425 1.761277 -0.220481 9 -0.640505 0.289374 -1.550670 >>> df.rolling(3).sum() A B C 0 NaN NaN NaN 1 NaN NaN NaN 2 -2.655105 0.637799 -2.135068 3 -0.971785 -0.600366 -3.280224 4 -0.214334 -1.294599 -3.227500 5 1.514216 2.028250 -2.989060 6 1.074618 5.709767 -2.322600 7 2.718061 3.850718 0.256446 8 -0.289082 2.454418 1.416871 9 0.212668 0.403198 -0.093924 >>> df.rolling(3).agg({'A':'sum', 'B':'min'}) A B 0 NaN NaN 1 NaN NaN 2 -2.655105 -0.165272 3 -0.971785 -1.340923 4 -0.214334 -1.340923 5 1.514216 -1.340923 6 1.074618 0.211596 7 2.718061 -1.647453 8 -0.289082 -1.647453 9 0.212668 -1.647453 RRRRRZsSeries/DataframeR RcO stt|j|||S(N(RRR(R+RRR,((s1lib/python2.7/site-packages/pandas/core/window.pyRsRsR RcC s,|jr|jddStt|jS(Nt roll_countR(R3RRRR(R+((s1lib/python2.7/site-packages/pandas/core/window.pyRs RcC s(tt|j|d|d|d|S(NRRR,(RRR(R+RRRR,((s1lib/python2.7/site-packages/pandas/core/window.pyRsRcO s,tjd||tt|j||S(NR(Rtvalidate_rolling_funcRRR(R+RR,((s1lib/python2.7/site-packages/pandas/core/window.pyRsRcO s,tjd||tt|j||S(NR(RR!RRR(R+RR,((s1lib/python2.7/site-packages/pandas/core/window.pyRsRcO s,tjd||tt|j||S(NR(RR!RRR(R+RR,((s1lib/python2.7/site-packages/pandas/core/window.pyRsRcO s,tjd||tt|j||S(NR(RR!RRR(R+RR,((s1lib/python2.7/site-packages/pandas/core/window.pyRsRcK stt|j|S(N(RRR(R+R,((s1lib/python2.7/site-packages/pandas/core/window.pyRsRicO s/tjd||tt|jd||S(NRR(RR!RRR(R+RRR,((s1lib/python2.7/site-packages/pandas/core/window.pyRsRcO s/tjd||tt|jd||S(NRR(RR!RRR(R+RRR,((s1lib/python2.7/site-packages/pandas/core/window.pyRsRcK stt|j|S(N(RRR(R+R,((s1lib/python2.7/site-packages/pandas/core/window.pyRssy Examples -------- The example below will show a rolling calculation with a window size of four matching the equivalent function call using `scipy.stats`. >>> arr = [1, 2, 3, 4, 999] >>> fmt = "{0:.6f}" # limit the printed precision to 6 digits >>> import scipy.stats >>> print(fmt.format(scipy.stats.kurtosis(arr[:-1], bias=False))) -1.200000 >>> print(fmt.format(scipy.stats.kurtosis(arr[1:], bias=False))) 3.999946 >>> s = pd.Series(arr) >>> s.rolling(4).kurt() 0 NaN 1 NaN 2 NaN 3 -1.200000 4 3.999946 dtype: float64 RcK stt|j|S(N(RRR(R+R,((s1lib/python2.7/site-packages/pandas/core/window.pyRsR R cK s"tt|jd|d||S(NR R (RRR (R+R R R,((s1lib/python2.7/site-packages/pandas/core/window.pyR sRcK s(tt|jd|d|d||S(NRTRR(RRR(R+RTRRR,((s1lib/python2.7/site-packages/pandas/core/window.pyRsRcK s"tt|jd|d||S(NRTR(RRR(R+RTRR,((s1lib/python2.7/site-packages/pandas/core/window.pyRsN((!RPRRR0R1R*RRRRRRRRRRRR'RRt _doc_templateRRRRRRRt_agg_docRR RR(((s1lib/python2.7/site-packages/pandas/core/window.pyR sp #  +                           tRollingGroupbycB s2eZdZedZddZdZRS(sS Provides a rolling groupby implementation. .. versionadded:: 0.18.1 cC stS(N(R(R+((s1lib/python2.7/site-packages/pandas/core/window.pyR/scC sX|jdk r9|jjj|j|j_d|_ntt|j||d|S(NRI( R!R'RR&t set_indexR1RR$RJ(R+RHR>RI((s1lib/python2.7/site-packages/pandas/core/window.pyRJs cC sdS(s Validate that on is monotonic; we don't care for groupby.rolling because we have already validated at a higher level. N((R+((s1lib/python2.7/site-packages/pandas/core/window.pyR!sN(RPRRRR/R'RJR(((s1lib/python2.7/site-packages/pandas/core/window.pyR$ s t Expandingc B s5eZdZdddgZdeddZedZd3dZ e d Z e d Z e d e d e d dddddeeddZeZe ddeeddZe ddeedd3d4idZe ddeeddZe ddeeeeddZe ddeeddZe ddeedd Ze ddeed!d"Ze ddeed#dd$Ze ddeed%dd&Ze ddeeeed'd(Ze d)Zeee ddeed*d+Ze ddeed,d-d.Ze ddeeeed/d3d3dd0Z e ddeed1d3d3d2Z!RS(5s Provides expanding transformations. .. versionadded:: 0.18.0 Parameters ---------- min_periods : int, default 1 Minimum number of observations in window required to have a value (otherwise result is NA). center : bool, default False Set the labels at the center of the window. axis : int or str, default 0 Returns ------- a Window sub-classed for the particular operation See Also -------- rolling : Provides rolling window calculations. ewm : Provides exponential weighted functions. Notes ----- By default, the result is set to the right edge of the window. This can be changed to the center of the window by setting ``center=True``. Examples -------- >>> df = pd.DataFrame({'B': [0, 1, 2, np.nan, 4]}) B 0 0.0 1 1.0 2 2.0 3 NaN 4 4.0 >>> df.expanding(2).sum() B 0 NaN 1 1.0 2 3.0 3 3.0 4 7.0 RRR iic K s/tt|jd|d|d|d|dS(NR&RRR (RR&R-(R+R&RRR R,((s1lib/python2.7/site-packages/pandas/core/window.pyR-^scC stS(N(R&(R+((s1lib/python2.7/site-packages/pandas/core/window.pyR/cscC sQ|jj|j}t||dk t|}|jpAd}t||S(su Get the window length over which to perform some operation. Parameters ---------- other : object, default None The other object that is involved in the operation. Such an object is involved for operations like covariance. Returns ------- window : int The window length. iN(R&t _get_axisR RR'RR(R+RTR tlength((s1lib/python2.7/site-packages/pandas/core/window.pyRUgs s See Also -------- pandas.DataFrame.expanding.aggregate pandas.DataFrame.rolling.aggregate pandas.DataFrame.aggregate s Examples -------- >>> df = pd.DataFrame(np.random.randn(10, 3), columns=['A', 'B', 'C']) >>> df A B C 0 -2.385977 -0.102758 0.438822 1 -1.004295 0.905829 -0.954544 2 0.735167 -0.165272 -1.619346 3 -0.702657 -1.340923 -0.706334 4 -0.246845 0.211596 -0.901819 5 2.463718 3.157577 -1.380906 6 -1.142255 2.340594 -0.039875 7 1.396598 -1.647453 1.677227 8 -0.543425 1.761277 -0.220481 9 -0.640505 0.289374 -1.550670 >>> df.ewm(alpha=0.5).mean() A B C 0 -2.385977 -0.102758 0.438822 1 -1.464856 0.569633 -0.490089 2 -0.207700 0.149687 -1.135379 3 -0.471677 -0.645305 -0.906555 4 -0.355635 -0.203033 -0.904111 5 1.076417 1.503943 -1.146293 6 -0.041654 1.925562 -0.588728 7 0.680292 0.132049 0.548693 8 0.067236 0.948257 0.163353 9 -0.286980 0.618493 -0.694496 RRRRRZsSeries/DataframeRcO stt|j|||S(N(RR&R(R+RRR,((s1lib/python2.7/site-packages/pandas/core/window.pyRsRst expandingRcK stt|j|S(N(RR&R(R+R,((s1lib/python2.7/site-packages/pandas/core/window.pyRsRcC s(tt|j|d|d|d|S(NRRR,(RR&R(R+RRRR,((s1lib/python2.7/site-packages/pandas/core/window.pyRsRcO s,tjd||tt|j||S(NR(Rtvalidate_expanding_funcRR&R(R+RR,((s1lib/python2.7/site-packages/pandas/core/window.pyRsRcO s,tjd||tt|j||S(NR(RR*RR&R(R+RR,((s1lib/python2.7/site-packages/pandas/core/window.pyRsRcO s,tjd||tt|j||S(NR(RR*RR&R(R+RR,((s1lib/python2.7/site-packages/pandas/core/window.pyRsRcO s,tjd||tt|j||S(NR(RR*RR&R(R+RR,((s1lib/python2.7/site-packages/pandas/core/window.pyRsRcK stt|j|S(N(RR&R(R+R,((s1lib/python2.7/site-packages/pandas/core/window.pyRsRcO s/tjd||tt|jd||S(NRR(RR*RR&R(R+RRR,((s1lib/python2.7/site-packages/pandas/core/window.pyRsRcO s/tjd||tt|jd||S(NRR(RR*RR&R(R+RRR,((s1lib/python2.7/site-packages/pandas/core/window.pyRsRcK stt|j|S(N(RR&R(R+R,((s1lib/python2.7/site-packages/pandas/core/window.pyRssz Examples -------- The example below will show an expanding calculation with a window size of four matching the equivalent function call using `scipy.stats`. >>> arr = [1, 2, 3, 4, 999] >>> import scipy.stats >>> fmt = "{0:.6f}" # limit the printed precision to 6 digits >>> print(fmt.format(scipy.stats.kurtosis(arr[:-1], bias=False))) -1.200000 >>> print(fmt.format(scipy.stats.kurtosis(arr, bias=False))) 4.999874 >>> s = pd.Series(arr) >>> s.expanding(4).kurt() 0 NaN 1 NaN 2 NaN 3 -1.200000 4 4.999874 dtype: float64 RcK stt|j|S(N(RR&R(R+R,((s1lib/python2.7/site-packages/pandas/core/window.pyRsR R cK s"tt|jd|d||S(NR R (RR&R (R+R R R,((s1lib/python2.7/site-packages/pandas/core/window.pyR sRcK s(tt|jd|d|d||S(NRTRR(RR&R(R+RTRRR,((s1lib/python2.7/site-packages/pandas/core/window.pyRsRcK s"tt|jd|d||S(NRTR(RR&R(R+RTRR,((s1lib/python2.7/site-packages/pandas/core/window.pyRsN(("RPRRR]RAR-RR/R'RURRRRRRRRRRRR"RRRRRRRR#RR RR(((s1lib/python2.7/site-packages/pandas/core/window.pyR&+sp/                             tExpandingGroupbycB seZdZedZRS(sU Provides a expanding groupby implementation. .. versionadded:: 0.18.1 cC stS(N(R&(R+((s1lib/python2.7/site-packages/pandas/core/window.pyR/$s(RPRRRR/(((s1lib/python2.7/site-packages/pandas/core/window.pyR+ss Parameters ---------- bias : bool, default False Use a standard estimation bias correction. *args, **kwargs Arguments and keyword arguments to be passed into func. s Parameters ---------- other : Series, DataFrame, or ndarray, optional If not supplied then will default to self and produce pairwise output. pairwise : bool, default None If False then only matching columns between self and other will be used and the output will be a DataFrame. If True then all pairwise combinations will be calculated and the output will be a MultiIndex DataFrame in the case of DataFrame inputs. In the case of missing elements, only complete pairwise observations will be used. bias : bool, default False Use a standard estimation bias correction. **kwargs Keyword arguments to be passed into func. tEWMc B seZdZdddddgZdddddeeddZedZ e d Z e d Z e d e d e d dddddeeddZeZdZe ddeedZe ddeeeeedZeZe ddeeeeedZe ddeeeeddedZe ddeeeedddZRS(s Provides exponential weighted functions. .. versionadded:: 0.18.0 Parameters ---------- com : float, optional Specify decay in terms of center of mass, :math:`\alpha = 1 / (1 + com),\text{ for } com \geq 0` span : float, optional Specify decay in terms of span, :math:`\alpha = 2 / (span + 1),\text{ for } span \geq 1` halflife : float, optional Specify decay in terms of half-life, :math:`\alpha = 1 - exp(log(0.5) / halflife),\text{ for } halflife > 0` alpha : float, optional Specify smoothing factor :math:`\alpha` directly, :math:`0 < \alpha \leq 1` .. versionadded:: 0.18.0 min_periods : int, default 0 Minimum number of observations in window required to have a value (otherwise result is NA). adjust : bool, default True Divide by decaying adjustment factor in beginning periods to account for imbalance in relative weightings (viewing EWMA as a moving average) ignore_na : bool, default False Ignore missing values when calculating weights; specify True to reproduce pre-0.15.0 behavior Returns ------- a Window sub-classed for the particular operation See Also -------- rolling : Provides rolling window calculations. expanding : Provides expanding transformations. Notes ----- Exactly one of center of mass, span, half-life, and alpha must be provided. Allowed values and relationship between the parameters are specified in the parameter descriptions above; see the link at the end of this section for a detailed explanation. When adjust is True (default), weighted averages are calculated using weights (1-alpha)**(n-1), (1-alpha)**(n-2), ..., 1-alpha, 1. When adjust is False, weighted averages are calculated recursively as: weighted_average[0] = arg[0]; weighted_average[i] = (1-alpha)*weighted_average[i-1] + alpha*arg[i]. When ignore_na is False (default), weights are based on absolute positions. For example, the weights of x and y used in calculating the final weighted average of [x, None, y] are (1-alpha)**2 and 1 (if adjust is True), and (1-alpha)**2 and alpha (if adjust is False). When ignore_na is True (reproducing pre-0.15.0 behavior), weights are based on relative positions. For example, the weights of x and y used in calculating the final weighted average of [x, None, y] are 1-alpha and 1 (if adjust is True), and 1-alpha and alpha (if adjust is False). More details can be found at http://pandas.pydata.org/pandas-docs/stable/computation.html#exponentially-weighted-windows Examples -------- >>> df = pd.DataFrame({'B': [0, 1, 2, np.nan, 4]}) B 0 0.0 1 1.0 2 2.0 3 NaN 4 4.0 >>> df.ewm(com=0.5).mean() B 0 0.000000 1 0.750000 2 1.615385 3 1.615385 4 3.670213 RRtadjustt ignore_naR ic C sR||_t|||||_||_||_||_| |_d|_dS(N( R&t_get_center_of_massRRR-R.R R'R!( R+R&RtspanthalflifetalphaRR-R.R ((s1lib/python2.7/site-packages/pandas/core/window.pyR-s     cC stS(N(R,(R+((s1lib/python2.7/site-packages/pandas/core/window.pyR/ssF See Also -------- pandas.DataFrame.rolling.aggregate s Examples -------- >>> df = pd.DataFrame(np.random.randn(10, 3), columns=['A', 'B', 'C']) >>> df A B C 0 -2.385977 -0.102758 0.438822 1 -1.004295 0.905829 -0.954544 2 0.735167 -0.165272 -1.619346 3 -0.702657 -1.340923 -0.706334 4 -0.246845 0.211596 -0.901819 5 2.463718 3.157577 -1.380906 6 -1.142255 2.340594 -0.039875 7 1.396598 -1.647453 1.677227 8 -0.543425 1.761277 -0.220481 9 -0.640505 0.289374 -1.550670 >>> df.ewm(alpha=0.5).mean() A B C 0 -2.385977 -0.102758 0.438822 1 -1.464856 0.569633 -0.490089 2 -0.207700 0.149687 -1.135379 3 -0.471677 -0.645305 -0.906555 4 -0.355635 -0.203033 -0.904111 5 1.076417 1.503943 -1.146293 6 -0.041654 1.925562 -0.588728 7 0.680292 0.132049 0.548693 8 0.067236 0.948257 0.163353 9 -0.286980 0.618493 -0.694496 RRRRRZsSeries/DataframeRcO stt|j|||S(N(RR,R(R+RRR,((s1lib/python2.7/site-packages/pandas/core/window.pyRsc  s)j\}}}g}x|D]}yj|j}Wn*tk rj|j|jjq"nX|jdkr|j|jq"nt|tj rt t |ddkrt djd|nfd}n|jtj|j|q"Wj|||S(s$ Rolling statistical measure using supplied function. Designed to be used with passed-in Cython array-based functions. Parameters ---------- func : str/callable to apply Returns ------- y : same type as input argument is3we do not support this function in libwindow.{func}Rc s4|jtjtjtjS(N(RRR-R.R(R(RR+(s1lib/python2.7/site-packages/pandas/core/window.pyR sN(RDRnRCRiRR=RRtRRR^RR'R8R_RjRR R( R+RR,R%R&R:RRRC((RR+s1lib/python2.7/site-packages/pandas/core/window.pyRs&     #RstewmcO s#tjd|||jd|S(s Exponential weighted moving average. Parameters ---------- *args, **kwargs Arguments and keyword arguments to be passed into func. Rtewma(RRR(R+RR,((s1lib/python2.7/site-packages/pandas/core/window.pyR s cO s,tjd||t|jd||S(s5 Exponential weighted moving stddev. Rtbias(RRRR(R+R5RR,((s1lib/python2.7/site-packages/pandas/core/window.pyR sc s5tjd||fd}j||S(s7 Exponential weighted moving variance. Rc  sCtj||jtjtjtjtS(N(RtewmcovRRR-R.R(R(R5R+(s1lib/python2.7/site-packages/pandas/core/window.pyR, s(RRR(R+R5RR,R((R5R+s1lib/python2.7/site-packages/pandas/core/window.pyR# sc ss|dkr0j}|dkr'tn|}nj|}fd}tj|j|dt|S(s9 Exponential weighted sample covariance. c  s|j|}j|}tj|j|jjtjtjtjt}|j |S(N( RERR6RnRRR-R.RR|(RRR(R5R+(s1lib/python2.7/site-packages/pandas/core/window.pyR@ sRN(R'R9RRERR(R+RTRR5R,R((R5R+s1lib/python2.7/site-packages/pandas/core/window.pyR3 s   c sp|dkr0j}|dkr'tn|}nj|}fd}tj|j|dt|S(s: Exponential weighted sample correlation. c  sj|}j|}fd}|j}|j}tjddF|||}|||}|||}|t||}WdQX|j|S(Nc s=tj||jtjtjtjdS(Ni(RR6RRR-R.R(Rty(R+(s1lib/python2.7/site-packages/pandas/core/window.pyt_cov] s  RR(RERnRjRRR|( RRR8tx_valuesty_valuesRtx_varty_varR(R+(s1lib/python2.7/site-packages/pandas/core/window.pyRY s  RN(R'R9RRERR(R+RTRR,R((R+s1lib/python2.7/site-packages/pandas/core/window.pyRL s  N(RPRRR]R'RRAR-RR/RRRRRRRRRR"Rt_bias_templateRtvolRt_pairwise_templateRR(((s1lib/python2.7/site-packages/pandas/core/window.pyR,FsJW       )             c sttjttfo3ttjttfsEtdnttjtfrttjtfrt\}}||Sttrddlmfd}i}ttr\|t krkrXxVt j D]E\}} j dd|fj dd|f|| s (&RtRjRuRRRiRJRvR@RAt enumerateR<RKt is_uniqueR8Rtcatch_warningsRt simplefiltertRuntimeWarningtaligntunionR:RtdictRFR}Rtnlevelst from_productRHtreorder_levelst sort_indextranget swaplevelt set_namestnamesR(RORPRRRRRCRRMRNt res_columnstk1tjtk2RFR}t result_indextcRz((R@RORPRs1lib/python2.7/site-packages/pandas/core/window.pyRr s   =   ! "   / [  !    cC sGtj||||}|dkr3tdn|dk r]|dkr=tdq=n|dk r|dkrtdn|dd}n|dk r|dkrtdndtjtjd|}d|d}nP|dk r1|dks|dkr td nd ||}n td t|S( Nis8comass, span, halflife, and alpha are mutually exclusiveis comass must satisfy: comass >= 0sspan must satisfy: span >= 1g@s#halflife must satisfy: halflife > 0g?s"alpha must satisfy: 0 < alpha <= 1g?s1Must pass one of comass, span, halflife, or alpha(Rtcount_not_noneR8R'RjtexptlogR(tcomassR0R1R2t valid_counttdecay((s1lib/python2.7/site-packages/pandas/core/window.pyR/ s*          cC set|st|}n|r/|ddnd}yt|SWntk r`|jtSXdS(Nig@i(R RRRiR(RRR((s1lib/python2.7/site-packages/pandas/core/window.pyR s  c sfd}|S(Nc s!|dkr|St|SdS(N(R'R(RR(tp(s1lib/python2.7/site-packages/pandas/core/window.pyt _check_func s ((RmRn((Rms1lib/python2.7/site-packages/pandas/core/window.pyR scC s|dkr|S|SdS(N(R'(RR((s1lib/python2.7/site-packages/pandas/core/window.pyR s cC stjdd tj|}|dk}WdQXt|trb|jjr{d||sb  @. H%W- t