ó šxŠ\c@sXddlmZddlZddlZddlZddlZddlZddlm Z m Z m Z ddl j Z ddlmZmZddlmZmZmZmZmZmZmZmZmZmZmZmZmZmZm Z m!Z!m"Z"ddl#m$Z$ddl%m&Z&m'Z'm(Z(ddl)j*j+Z,ddl-m.Z.e/Z0d Z1y\ddl2Z3e3j4Z5ee5ƒee1ƒkZ0e0s²ej6d j7d e5d e1ƒe8ƒnWne9k rÆnXe/a:e;d „Z<e<e.dƒƒde=fd„ƒYZ>de=fd„ƒYZ?d„Z@d„ZAddd„ZCdde/e;dd„ZDd„ZEd„ZFdd„ZGd„ZHde;dd„ZIde;dd„ZJe>dƒde;ddd„ƒZKe>de$ƒe?ƒde;dd „ƒƒZLe>dƒe?ƒde;dd!„ƒƒZMeNd"„ZOe>dƒe?d#d$ƒde;d$dd%„ƒƒZPe>dƒe?d#d$ƒde;d$dd&„ƒƒZQe>dd'ƒde;d$dd(„ƒZRd)„ZSeSd*d+d,ƒZTeSd-d+d.ƒZUe>d/ƒde;dd0„ƒZVe>d/ƒde;dd1„ƒZWe>dd'ƒde;dd2„ƒZXe>dd'ƒde;dd3„ƒZYe>dd'ƒde;ddd4„ƒZZd5„Z[eNd6„Z\d$d7„Z]d8„Z^e>dd'ƒd9dd:„ƒZ_d;„Z`e>dd'ƒdd<„ƒZad=„Zbd>„ZcecejdƒZeecejfƒZgecejhƒZiecejjƒZkecejlƒZmecejnƒZod?„Zpd@„ZqdS(Aiÿÿÿÿ(t LooseVersionN(tiNaTtlibttslibs(t _int64_maxtmaybe_upcast_putmask(t _get_dtypetis_any_int_dtypet is_bool_dtypet is_complextis_complex_dtypetis_datetime64_dtypetis_datetime64tz_dtypetis_datetime_or_timedelta_dtypetis_floattis_float_dtypet is_integertis_integer_dtypetis_numeric_dtypetis_object_dtypet is_scalartis_timedelta64_dtypet pandas_dtype(tDatetimeTZDtype(tisnatna_value_for_dtypetnotna(t get_options1.0.0sˆThe installed version of bottleneck {ver} is not supported in pandas and will be not be used The minimum supported version is {min_ver} tvertmin_vercCstr|andS(N(t_BOTTLENECK_INSTALLEDt_USE_BOTTLENECK(tv((s1lib/python2.7/site-packages/pandas/core/nanops.pytset_use_bottleneck/sscompute.use_bottlenecktdisallowcBs#eZd„Zd„Zd„ZRS(cGs0tt|ƒjƒtd„|Dƒƒ|_dS(Ncss|]}t|ƒjVqdS(N(Rttype(t.0tdtype((s1lib/python2.7/site-packages/pandas/core/nanops.pys =s(tsuperR"t__init__ttupletdtypes(tselfR)((s1lib/python2.7/site-packages/pandas/core/nanops.pyR';scCs%t|dƒo$t|jj|jƒS(NR%(thasattrt issubclassR%R#R)(R*tobj((s1lib/python2.7/site-packages/pandas/core/nanops.pytcheck?scs%tjˆƒ‡‡fd†ƒ}|S(NcsÍtj|tj|ƒƒ}t‡fd†|Dƒƒrgd}t|jdˆjjddƒƒƒ‚ny*t j ddƒˆ||ŽSWdQXWn5t k rÈ}t |dƒrÂt|ƒ‚n‚nXdS( Nc3s|]}ˆj|ƒVqdS(N(R.(R$R-(R*(s1lib/python2.7/site-packages/pandas/core/nanops.pys Gss7reduction operation {name!r} not allowed for this dtypetnametnanttinvalidtignorei( t itertoolstchaintcompatt itervaluestanyt TypeErrortformatt__name__treplacetnpterrstatet ValueErrorR(targstkwargstobj_itertmsgte(tfR*(s1lib/python2.7/site-packages/pandas/core/nanops.pyt_fDs*(t functoolstwraps(R*RERF((RER*s1lib/python2.7/site-packages/pandas/core/nanops.pyt__call__Cs!(R;t __module__R'R.RI(((s1lib/python2.7/site-packages/pandas/core/nanops.pyR"9s  tbottleneck_switchcBseZd„Zd„ZRS(cKs ||_dS(N(RA(R*RA((s1lib/python2.7/site-packages/pandas/core/nanops.pyR'[scsmˆj‰yttˆƒ‰Wnttfk r;d‰nXtjˆƒdt‡‡‡‡fd†ƒ}|S(NcsotˆjƒdkrTx<tjˆjƒD]%\}}||kr(|||t|ƒ r:t|ƒp"t|ƒ r:|dkr6tStStS(Ntnansumtnanprod(RdRe(RR R tFalseRc(tdtR/((s1lib/python2.7/site-packages/pandas/core/nanops.pyRU’s   cCst|tjƒrY|jdkr4tj|jƒƒS|jdkrYtj|jƒƒSnytj|ƒj ƒSWnt t fk rŠt SXdS(Ntf8tf4( t isinstanceR=tndarrayR%Rt has_infs_f8travelt has_infs_f4tisinfR8R9tNotImplementedErrorRf(R[((s1lib/python2.7/site-packages/pandas/core/nanops.pyRV§scCs{|dk r|St|ƒrM|dkr/tjS|dkrBtjStj Sn*|dkr`tjS|dkrptStjSdS(s; return the correct fill value for the dtype of the values s+infN(RSt _na_ok_dtypeR=R0tinfRRR(R%t fill_valuetfill_value_typ((s1lib/python2.7/site-packages/pandas/core/nanops.pyt_get_fill_value´s       c Cs‘t|ƒr*|j}t|d|ƒ}ntj|ƒ}|j}|dkrr|rct|ƒ}qrt|ƒ}nt|ƒsŠt|ƒr±t|d|ƒ}|j t j ƒ}nt |ƒ}t |d|d|ƒ}|r'|rð|jƒ}n|r t j|||ƒq<t|||ƒ\}} n|r<|jƒ}n|} t|ƒsZt|ƒrft j } nt|ƒr~t j} n|||| |fS(s“ utility to get the values view, mask, dtype if necessary copy and mask using the specified fill_value copy = True will force the copy t_valuestasi8RsRtN(R R%R_tcomtvalues_from_objectRSt _isfiniteRR tviewR=tint64RqRutcopytputmaskRRRRtfloat64( RXRNRsRttisfiniteR}tmaskR%tdtype_oktchangedt dtype_max((s1lib/python2.7/site-packages/pandas/core/nanops.pyt _get_valuesËs:         cCskt|ƒrt|ƒSt|ƒsFt|ƒsFt|ƒsFt|ƒrTtj|ƒStj|jdƒƒS(NR( R RR RRRR=R€tastype(RX((s1lib/python2.7/site-packages/pandas/core/nanops.pyRzs   cCs#t|jtjtjtjfƒ S(N(R,R#R=tintegert timedelta64t datetime64(R%((s1lib/python2.7/site-packages/pandas/core/nanops.pyRqs cCs=t|ƒst|ƒr¬|dkr-t}nt|tjƒsšt|ddƒ}t|ƒ sjt dƒ‚||kr‚tj }nt j |d|ƒ}q9|j |ƒ}nt|ƒr9t|tjƒs||krâtj }ntj|ƒtkrtdƒ‚nt j|ddƒ}q9|jdƒj |ƒ}n|S(s wrap our results if needed ttzsExpected non-null fill_valuesoverflow in timedelta operationtunittnsti8N(R R RSRRjR=RkR_RtAssertionErrorR0Rt TimestampR{RtfabsRR?t TimedeltaR†(R[R%RsRŠ((s1lib/python2.7/site-packages/pandas/core/nanops.pyt _wrap_resultss&       cCs‰t|ƒr|jdƒ}nt|jƒ}|jdkr@|S|j| |j|d}tj|d|jƒ}|j|ƒ|SdS(s]Return the missing value for `values` Parameters ---------- values : ndarray axis : int or None axis for the reduction Returns ------- result : scalar or ndarray For 1-D values, returns a scalar of the correct missing type. For 2-D values, returns a 1-D array where each element is missing. RiR%N( RR†RR%tndimtshapeR=temptytfill(RXRMRst result_shapeR[((s1lib/python2.7/site-packages/pandas/core/nanops.pyRT3s   cCs:t||td|d|ƒ\}}}}}|j|ƒS(s  Check if any elements along an axis evaluate to True. Parameters ---------- values : ndarray axis : int, optional skipna : bool, default True mask : ndarray[bool], optional nan-mask if known Returns ------- result : bool Examples -------- >>> import pandas.core.nanops as nanops >>> s = pd.Series([1, 2]) >>> nanops.nanany(s) True >>> import pandas.core.nanops as nanops >>> s = pd.Series([np.nan]) >>> nanops.nanany(s) False R}R(R…RfR8(RXRMRNRR%t_((s1lib/python2.7/site-packages/pandas/core/nanops.pytnananyQscCs:t||td|d|ƒ\}}}}}|j|ƒS(s Check if all elements along an axis evaluate to True. Parameters ---------- values : ndarray axis: int, optional skipna : bool, default True mask : ndarray[bool], optional nan-mask if known Returns ------- result : bool Examples -------- >>> import pandas.core.nanops as nanops >>> s = pd.Series([1, 2, np.nan]) >>> nanops.nanall(s) True >>> import pandas.core.nanops as nanops >>> s = pd.Series([1, 0]) >>> nanops.nanall(s) False R}R(R…Rctall(RXRMRNRR%R˜((s1lib/python2.7/site-packages/pandas/core/nanops.pytnanallrstM8ic Cs”t||dd|ƒ\}}}}}|}t|ƒrB|}nt|ƒrZtj}n|j|d|ƒ} t| ||d|ƒ} t| |ƒS(s¾ Sum the elements along an axis ignoring NaNs Parameters ---------- values : ndarray[dtype] axis: int, optional skipna : bool, default True min_count: int, default 0 mask : ndarray[bool], optional nan-mask if known Returns ------- result : dtype Examples -------- >>> import pandas.core.nanops as nanops >>> s = pd.Series([1, 2, np.nan]) >>> nanops.nansum(s) 3.0 iRR%RL(R…RRR=Rtsumt_maybe_null_outR’( RXRMRNRLRR%R„R˜t dtype_sumtthe_sum((s1lib/python2.7/site-packages/pandas/core/nanops.pyRd“s!    c CsUt||dd|ƒ\}}}}}|}tj}t|ƒsft|ƒsft|ƒsft|ƒrrtj}nt|ƒr|}|}nt||d|ƒ} t |j |d|ƒƒ} |dk r)t | dt ƒr)tjddƒ| | } WdQX| dk} | jƒrHtj| | >> import pandas.core.nanops as nanops >>> s = pd.Series([1, 2, np.nan]) >>> nanops.nanmean(s) 1.5 iRR%R“RšR3N(R…R=RRRR R Rt _get_countst_ensure_numericRRSR_RfR>R8R0R’( RXRMRNRR%R„R˜RŸt dtype_counttcountR tthe_meantct_mask((s1lib/python2.7/site-packages/pandas/core/nanops.pytnanmean¹s($      c sN‡fd†}t|ˆd|ƒ\}}}}}t|ƒs^|jdƒ}tj||>> import pandas.core.nanops as nanops >>> s = pd.Series([1, np.nan, 2, 2]) >>> nanops.nanmedian(s) 2.0 cs8t|ƒ}ˆ r'|jƒ r'tjStj||ƒS(N(RRšR=R0t nanmedian(txR(RN(s1lib/python2.7/site-packages/pandas/core/nanops.pyt get_medians RRhiN(R…RR†R=R0RSRmRQR“R’tapply_along_axisR¨tarrayR”tarangeR•R–( RXRMRNRRªR%R„R˜tnotemptytshptdimstret((RNs1lib/python2.7/site-packages/pandas/core/nanops.pyR¨îs($    cCsµt|ƒ}t||d|ƒ}||j|ƒ}t|ƒrd||kr«tj}tj}q«nG||k}|jƒr«tj||tjƒtj||tjƒn||fS(NR%(RR¡R#RR=R0R8R~(RRMtddofR%R¤tdtmask2((s1lib/python2.7/site-packages/pandas/core/nanops.pyt_get_counts_nanvar1s      R²ic Cs=tjt|d|d|d|d|ƒƒ}t||jƒS(s¸ Compute the standard deviation along given axis while ignoring NaNs Parameters ---------- values : ndarray axis: int, optional skipna : bool, default True ddof : int, default 1 Delta Degrees of Freedom. The divisor used in calculations is N - ddof, where N represents the number of elements. mask : ndarray[bool], optional nan-mask if known Returns ------- result : float Unless input is a float array, in which case use the same precision as the input array. Examples -------- >>> import pandas.core.nanops as nanops >>> s = pd.Series([1, np.nan, 2, 3]) >>> nanops.nanstd(s) 1.0 RMRNR²R(R=tsqrttnanvarR’R%(RXRMRNR²RR[((s1lib/python2.7/site-packages/pandas/core/nanops.pytnanstdCs! c Cs…tj|ƒ}|j}|dkr3t|ƒ}nt|ƒr^|jdƒ}tj||>> import pandas.core.nanops as nanops >>> s = pd.Series([1, np.nan, 2, 3]) >>> nanops.nanvar(s) 1.0 RhiRMR%iN(RxRyR%RSRRR†R=R0RRµR}R~R¢RRt expand_dimsR’( RXRMRNR²RR%R¤R³tavgtsqrR[((s1lib/python2.7/site-packages/pandas/core/nanops.pyR·fs,    ! %  tm8cCs¨t|||d|d|ƒ|dkr7t|ƒ}nt|jƒsX|jdƒ}nt||||jƒ\}}t|||d|ƒ}tj|ƒtj|ƒS(sÒ Compute the standard error in the mean along given axis while ignoring NaNs Parameters ---------- values : ndarray axis: int, optional skipna : bool, default True ddof : int, default 1 Delta Degrees of Freedom. The divisor used in calculations is N - ddof, where N represents the number of elements. mask : ndarray[bool], optional nan-mask if known Returns ------- result : float64 Unless input is a float array, in which case use the same precision as the input array. Examples -------- >>> import pandas.core.nanops as nanops >>> s = pd.Series([1, np.nan, 2, 3]) >>> nanops.nansem(s) 0.5773502691896258 R²RRhN( R·RSRRR%R†RµR=R¶(RXRMRNR²RR¤R˜tvar((s1lib/python2.7/site-packages/pandas/core/nanops.pytnansemªs  cs5tƒdtd‡‡fd†ƒ}dˆ|_|S(Ncsót||dˆd|ƒ\}}}}}|dk rI|j|dksX|jdkr¼y/t|ˆƒ|d|ƒ}|jtjƒWqÑtt t tj j j fk r¸tj}qÑXnt|ˆƒ|ƒ}t|||ƒ}t|||ƒS(NRtRiR%(R…RSR”RQR_R–R=R0RaR9R?tcoret _internalt AxisErrorR’Rž(RXRMRNRR%R„RsR[(Rttmeth(s1lib/python2.7/site-packages/pandas/core/nanops.pyt reduction×s'R0(RKRSRcR;(RÂRtRÃ((RtRÂs1lib/python2.7/site-packages/pandas/core/nanops.pyt _nanminmaxÖs tminRts+inftmaxs-inftOcCsRt||ddd|ƒ\}}}}}|j|ƒ}t||||ƒ}|S(s³ Parameters ---------- values : ndarray axis: int, optional skipna : bool, default True mask : ndarray[bool], optional nan-mask if known Returns -------- result : int The index of max value in specified axis or -1 in the NA case Examples -------- >>> import pandas.core.nanops as nanops >>> s = pd.Series([1, 2, 3, np.nan, 4]) >>> nanops.nanargmax(s) 4 Rts-infR(R…targmaxt_maybe_arg_null_out(RXRMRNRR%R˜R[((s1lib/python2.7/site-packages/pandas/core/nanops.pyt nanargmaxós 'cCsRt||ddd|ƒ\}}}}}|j|ƒ}t||||ƒ}|S(s³ Parameters ---------- values : ndarray axis: int, optional skipna : bool, default True mask : ndarray[bool], optional nan-mask if known Returns -------- result : int The index of min value in specified axis or -1 in the NA case Examples -------- >>> import pandas.core.nanops as nanops >>> s = pd.Series([1, 2, 3, np.nan, 4]) >>> nanops.nanargmin(s) 0 Rts+infR(R…targminRÉ(RXRMRNRR%R˜R[((s1lib/python2.7/site-packages/pandas/core/nanops.pyt nanargmins 'c Cs<tj|ƒ}|d kr*t|ƒ}nt|jƒsZ|jdƒ}t||ƒ}nt||d|jƒ}|rš|jƒ}t j ||dƒn|j |dt j ƒ|}|d k r×t j ||ƒ}n||}|rýt j ||dƒn|d}||}|j |dt j ƒ} |j |dt j ƒ} t| ƒ} t| ƒ} t jddddƒ+||dd |d| | d } Wd QX|j} t| ƒrÅ| j| ƒ} nt| t jƒr t j| dkd| ƒ} t j| |d k<| S| dkrdn| } |d kr4t jS| Sd S( sÈ Compute the sample skewness. The statistic computed here is the adjusted Fisher-Pearson standardized moment coefficient G1. The algorithm computes this coefficient directly from the second and third central moment. Parameters ---------- values : ndarray axis: int, optional skipna : bool, default True mask : ndarray[bool], optional nan-mask if known Returns ------- result : float64 Unless input is a float array, in which case use the same precision as the input array. Examples -------- >>> import pandas.core.nanops as nanops >>> s = pd.Series([1,np.nan, 1, 2]) >>> nanops.nanskew(s) 1.7320508075688787 RhR%iiR2R3tdivideigà?gø?Ni(RxRyRSRRR%R†R¡R}R=R~RRR¹t_zero_out_fperrR>RjRktwhereR0( RXRMRNRR¤tmeantadjustedt adjusted2t adjusted3tm2tm3R[R%((s1lib/python2.7/site-packages/pandas/core/nanops.pytnanskew/sF        ,   cCs¡tj|ƒ}|d kr*t|ƒ}nt|jƒsZ|jdƒ}t||ƒ}nt||d|jƒ}|rš|jƒ}t j ||dƒn|j |dt j ƒ|}|d k r×t j ||ƒ}n||}|rýt j ||dƒn|d}|d}|j |dt j ƒ} |j |dt j ƒ} t jddddƒ[d|d d|d|d} ||d |d | } |d|d| d} Wd QXt| ƒ} t| ƒ} t| t jƒs|d krót jS| dkrdSnt jddddƒ| | | }Wd QX|j}t|ƒrZ|j|ƒ}nt|t jƒrt j| dkd|ƒ}t j||d k>> import pandas.core.nanops as nanops >>> s = pd.Series([1,np.nan, 1, 3, 2]) >>> nanops.nankurt(s) -1.2892561983471076 RhR%iiR2R3RÍiiNi(RxRyRSRRR%R†R¡R}R=R~RRR¹R>RÎRjRkR0RÏ(RXRMRNRR¤RÐRÑRÒt adjusted4RÔtm4tadjtnumertdenomR[R%((s1lib/python2.7/site-packages/pandas/core/nanops.pytnankurt~sP      "       cCsl|dkrt|ƒ}n|rGt|ƒ rG|jƒ}d||>> import pandas.core.nanops as nanops >>> s = pd.Series([1, 2, 3, np.nan]) >>> nanops.nanprod(s) 6.0 Returns -------- The product of all elements on a given axis. ( NaNs are treated as 1) iRLN(RSRRR}tprodRž(RXRMRNRLRR[((s1lib/python2.7/site-packages/pandas/core/nanops.pyRe×s   cCs™|dkst|dtƒ rU|r=|jƒrRd}qRq•|jƒr•d}q•n@|rm|j|ƒ}n|j|ƒ}|jƒr•d||RÏtabsR%R#(targ((s1lib/python2.7/site-packages/pandas/core/nanops.pyRÎ1s+tpearsoncCs§t|ƒt|ƒkr'tdƒ‚n|dkr<d}nt|ƒt|ƒ@}|jƒsu||}||}nt|ƒ|krŽtjSt|ƒ}|||ƒS(s a, b: ndarrays s'Operands to nancorr must have same sizeiN(RORŽRSRRšR=R0t get_corr_func(tatbtmethodt min_periodstvalidRE((s1lib/python2.7/site-packages/pandas/core/nanops.pytnancorr:s      cs|d kr%ddlm‰m‰nt|ƒr5|Sd„}‡fd†}‡fd†}i|d6|d6|d6}||S( Ntkendalltspearmaniÿÿÿÿ(t kendalltaut spearmanrcSstj||ƒdS(Nii(ii(R=tcorrcoef(RçRè((s1lib/python2.7/site-packages/pandas/core/nanops.pyt_pearsonWscs*ˆ||ƒ}t|tƒr&|dS|S(Ni(RjR((RçRètrs(Rï(s1lib/python2.7/site-packages/pandas/core/nanops.pyt_kendallZscsˆ||ƒdS(Ni((RçRè(Rð(s1lib/python2.7/site-packages/pandas/core/nanops.pyt _spearman`sRå(RíRî(t scipy.statsRïRðtcallable(RéRòRôRõt _cor_methods((RïRðs1lib/python2.7/site-packages/pandas/core/nanops.pyRæQs    cCs¢t|ƒt|ƒkr'tdƒ‚n|dkr<d}nt|ƒt|ƒ@}|jƒsu||}||}nt|ƒ|krŽtjStj||ƒdS(Ns&Operands to nancov must have same sizeii(ii(RORŽRSRRšR=R0tcov(RçRèRêRë((s1lib/python2.7/site-packages/pandas/core/nanops.pytnancovks     cCs<t|tjƒr±t|ƒs*t|ƒr?|jtjƒ}q8t|ƒr8y|jtjƒ}Wn)t t fk rŒ|jtjƒ}q®Xtj |j ƒs®|j }q®q8n‡t|ƒpÒt|ƒpÒt|ƒs8yt|ƒ}Wq8tk r4yt|ƒ}Wq5tk r0t djd|ƒƒ‚q5Xq8Xn|S(Ns&Could not convert {value!s} to numerictvalue(RjR=RkRRR†RRt complex128R9R?R8timagtrealRRR tfloatRWtcomplexR:(R©((s1lib/python2.7/site-packages/pandas/core/nanops.pyR¢~s( $   cs‡fd†}|S(Ncs‘t|ƒ}t|ƒ}||B}tjddƒˆ||ƒ}WdQX|jƒrt|ƒrt|jdƒ}ntj||tjƒn|S(NRšR3RÇ(RR=R>R8RR†R~R0(R©tytxmasktymaskRR[(top(s1lib/python2.7/site-packages/pandas/core/nanops.pyRE™s     ((RRE((Rs1lib/python2.7/site-packages/pandas/core/nanops.pyt make_nancomp˜scCsl||}t|ƒdkrVtj|ƒr0|Stj|gt|ƒd|jƒSntj||d|ƒS(sÚ Wraper for np.percentile that skips missing values, specialized to 1-dimensional case. Parameters ---------- values : array over which to find quantiles mask : ndarray[bool] locations in values that should be considered missing q : scalar or array of quantile indices to find na_value : scalar value to return for empty or all-null values interpolation : str Returns ------- quantiles : scalar or array iR%t interpolation(RORRR=R¬R%t percentile(RXRtqtna_valueR((s1lib/python2.7/site-packages/pandas/core/nanops.pyt_nanpercentile_1d³s  c Cstj|ƒ rú|jƒrú|dkrAt||||d|ƒS|j|jkrh|j|jƒ}n|dkr‰|j}|j}ngtt |ƒt |ƒƒD]'\}}t||||d|ƒ^q¥} t j | d|j dt ƒj} | Snt j||d|d|ƒSdS(s× Wraper for np.percentile that skips missing values. Parameters ---------- values : array over which to find quantiles q : scalar or array of quantile indices to find axis : {0, 1} na_value : scalar value to return for empty or all-null values mask : ndarray[bool] locations in values that should be considered missing ndim : {1, 2} interpolation : str Returns ------- quantiles : scalar or array iRiR%R}RMN(RRR8R R“treshapeR”tTtziptlistR=R¬R%RfR( RXRRMR RR“RtvaltmR[((s1lib/python2.7/site-packages/pandas/core/nanops.pyt nanpercentileÓs    F!(rtdistutils.versionRRGR4toperatortwarningstnumpyR=t pandas._libsRRRt pandas.compatR6tpandas.core.dtypes.castRRtpandas.core.dtypes.commonRRRR R R R R RRRRRRRRRtpandas.core.dtypes.dtypesRtpandas.core.dtypes.missingRRRtpandas.core.commonR¿tcommonRxtpandas.core.configRRfRt_MIN_BOTTLENECK_VERSIONt bottleneckR`t __version__RtwarnR:t UserWarningt ImportErrorRRcR!tobjectR"RKRURVRSRuR…RzRqR’RTR™R›RdR§R¨RÿRµR¸R·R¾RÄtnanmintnanmaxRÊRÌRÖRÜReRÉR¡RžRÎRìRæRúR¢RtgttnangttgetnangetlttnanlttletnanleteqtnaneqtnetnanneR R(((s1lib/python2.7/site-packages/pandas/core/nanops.pyts¶     p       9  9   !! % 3 A   !  B +    N X #