W[c@ sdZddlmZddlZddlZddlmZddlj Z ddlm Z ddl Z ddlZddlmZddlmZdd lmZdd lmZmZmZmZdd lmZd d gZdZdZdZ dZ!de"fdYZ#e$e$e$e$e%e$de$dde&e$e$e%dde$e$dZ'de"fdYZ(e$de&dde%e$dZ)defdYZ*e$dde$e$e$e$e&e&e$e$e$e$e$d Z+dS(!s(Functions to visualize matrices of data.i(tdivisionN(tLineCollection(tgridspec(t hierarchyi(tcm(tGrid(tdespinetaxis_ticklabels_overlaptrelative_luminancetto_utf8(t string_typestheatmapt clustermapcC s6t|tjr+djtt|jS|jSdS(s6Convert a pandas index or multiindex to an axis label.t-N(t isinstancetpdt MultiIndextjointmapR tnamestname(tindex((s-lib/python2.7/site-packages/seaborn/matrix.pyt_index_to_labelscC sIt|tjr>g|jD]}djtt|^qS|jSdS(s5Convert a pandas index or multiindex into ticklabels.R N(RRRtvaluesRRR (Rti((s-lib/python2.7/site-packages/seaborn/matrix.pyt_index_to_ticklabels s,c stjjjttjrAtjfdDSttjr`jSy%dt tSWn4t k rgD]}t t|^qSXdS(sAConvert either a list of colors or nested lists of colors to RGB.c s&i|]}|j|qS((R(t.0tcol(tcolorstto_rgb(s-lib/python2.7/site-packages/seaborn/matrix.pys .s iN( tmplRtcolorConverterRRRt DataFrametSeriesRtlistt ValueError(Rtl((RRs-lib/python2.7/site-packages/seaborn/matrix.pyt_convert_colors(s   cC s|dkr'tj|jtj}nt|tjr|j|jkrZtdntj |d|j d|j dtj}nUt|tj r|j j |j  r|j j |j rd}t|qn|tj |B}|S(sEnsure that data and mask are compatabile and add missing values. Values will be plotted for cells where ``mask`` is ``False``. ``data`` is expected to be a DataFrame; ``mask`` can be an array or a DataFrame. s&Mask must have the same shape as data.Rtcolumnstdtypes2Mask must have the same index and columns as data.N(tNonetnptzerostshapetboolRtndarrayR#RR RR&tequalstisnull(tdatatmaskterr((s-lib/python2.7/site-packages/seaborn/matrix.pyt _matrix_mask<s    t _HeatMappercB sMeZdZeeddZdZdZdZdZ dZ RS(s?Draw a heatmap plot of a matrix with nice labels and colormaps.cC st|tjr|j}ntj|}tj|}t||}tjjtj||}d}t| t r| }t |j } n3| t krt |j } n| t krg} nd}t| t r| }t |j} n3| t krt |j} n| t kr)g} n|jj\}}t| s\g|_g|_nWt| tr| dkrd|_t |j |_n|j| |\|_|_t| sg|_g|_nWt| tr | dkr d|_t |j|_n|j| |\|_|_t|j }t|j}|dk r[|nd|_|dk rv|nd|_|j|||||||dkrt }d}nut|tr|r|}q+d}nNy |j}Wntk r|}nX|j|jkr%tdnt }||_ ||_!||_"||_#||_$| dkrjin| |_%| |_&| dkrin| |_'|j'j(dt)j*j+ddS(sInitialize the plotting object.itautotsDData supplied to "annot" must be the same shape as the data to plot.tticksiN(,RRR RR)tasarrayR3tmat masked_wheretintRR&tTruetFalseRtTR+tlentxtickst xticklabelsR t _skip_tickstytickst yticklabelsRR(txlabeltylabelt_determine_cmap_paramsR,tAttributeErrorR#R0t plot_datatannott annot_datatfmtt annot_kwstcbartcbar_kwst setdefaultRttickert MaxNLocator(tselfR0tvmintvmaxtcmaptcentertrobustRJRLRMRNRORARDR1RIt xtickeveryt ytickeverytnxtnyRERFRK((s-lib/python2.7/site-packages/seaborn/matrix.pyt__init__ds                               c C s|jtj|j}|dkrM|r>tj|dn |j}n|dkr|rqtj|dn |j}n|||_|_|dkr|dkrt j |_ q t j |_ nWt |trtj j||_ n0t |trtjj||_ n ||_ |dk rt||||}tjj||||} | ||g\} } tj| | d} tjj|j | |_ ndS(s@Use some heuristics to set good defaults for colorbar and range.iibiN(R0R)tisnanR(t percentiletmintmaxRTRURtrocketRVticefireRR Rtget_cmapR"RtListedColormapt Normalizetlinspace( RSRIRTRURVRWRXt calc_datatvrangetnormlizetcmintcmaxtcc((s-lib/python2.7/site-packages/seaborn/matrix.pyRGs* ' '    cC s0|j|jj\}}tjtj|dtj|d\}}xt|j|j|j|j |jjD]\}}} } } | tj j k rt | } | dkrdnd} d|j dj| }td| dd d d }|j|j|j||||qqWd S( s/Add textual labels with the value in each cell.g?g&1?s.15tws{:t}tcolorthaRWtvaN(tupdate_scalarmappableRKR+R)tmeshgridtarangetziptflatt get_arraytget_facecolorsR9tmaskedRRLtformattdicttupdateRMttext(RStaxtmeshtheighttwidthtxpostypostxtytmRptvaltlumt text_colort annotationt text_kwargs((s-lib/python2.7/site-packages/seaborn/matrix.pyt_annotate_heatmaps 2" cC st|}|dkr(gg}}ni|dkrQtj|d|}}n@d||}}}tj|||d}||||}||fS(s3Return ticks and labels at evenly spaced intervals.iig?(R?R)Ru(RStlabelst tickeverytnR7tstarttendtstep((s-lib/python2.7/site-packages/seaborn/matrix.pyRBs   c C s|jjj}|jj|}|j|jg|}|j|jg|}|j dg\}|j j }t ||d} | dkrggfSt || d} | dkrdn| } |j|| \} }| |fS(s5Determine ticks and ticklabels that minimize overlap.iiHi(tfiguretdpi_scale_transtinvertedtget_window_extentt transformedRRtxaxistyaxist set_tickstlabeltget_sizeR;R?RB( RSRRtaxist transformtbboxtsizetticktfontsizet max_tickst tick_everyR7((s-lib/python2.7/site-packages/seaborn/matrix.pyt _auto_tickss  c C std|dtdt|j|jd|jd|jd|j|}|jdd|jj d fd d|jj df|j |j r|j j ||||j}|jjd|jd tr|jjtqnt|jtr*|jd kr*|j||jd\}}n|j|j}}t|jtr|jd kr|j||jd \}} n|j|j}} |jd |d||j|} |j| dd} t|j jdr|j j |j jj!nt"| r"t#j$| ddnt"| rDt#j$| ddn|jd|j%d|j&|j'r||j(||ndS(s&Draw the heatmap on the provided Axes.RtlefttbottomRTRURVtxlimiitylimt rasterizedR5R@RCtrotationtverticalt get_renderert horizontalRERFN()RR<t pcolormeshRIRTRURVtsetR0R+t invert_yaxisRNRtcolorbarROtoutlinet set_linewidthtgetR=tsolidstset_rasterizedRR@R RRARCRDtset_xticklabelstset_yticklabelsthasattrtcanvastdrawRRtplttsetpRERFRJR( RSRtcaxtkwsRtcbR@RARCRDtxtltytl((s-lib/python2.7/site-packages/seaborn/matrix.pytplots8!6  !!!!   N( t__name__t __module__t__doc__R<R(R]RGRRBRR(((s-lib/python2.7/site-packages/seaborn/matrix.pyR4asc   s.2gitwhiteR5cK st|||||||||| | |||}| |d<| |d<|dkrbtj}n|rx|jdn|j|| ||S(sRPlot rectangular data as a color-encoded matrix. This is an Axes-level function and will draw the heatmap into the currently-active Axes if none is provided to the ``ax`` argument. Part of this Axes space will be taken and used to plot a colormap, unless ``cbar`` is False or a separate Axes is provided to ``cbar_ax``. Parameters ---------- data : rectangular dataset 2D dataset that can be coerced into an ndarray. If a Pandas DataFrame is provided, the index/column information will be used to label the columns and rows. vmin, vmax : floats, optional Values to anchor the colormap, otherwise they are inferred from the data and other keyword arguments. cmap : matplotlib colormap name or object, or list of colors, optional The mapping from data values to color space. If not provided, the default will depend on whether ``center`` is set. center : float, optional The value at which to center the colormap when plotting divergant data. Using this parameter will change the default ``cmap`` if none is specified. robust : bool, optional If True and ``vmin`` or ``vmax`` are absent, the colormap range is computed with robust quantiles instead of the extreme values. annot : bool or rectangular dataset, optional If True, write the data value in each cell. If an array-like with the same shape as ``data``, then use this to annotate the heatmap instead of the raw data. fmt : string, optional String formatting code to use when adding annotations. annot_kws : dict of key, value mappings, optional Keyword arguments for ``ax.text`` when ``annot`` is True. linewidths : float, optional Width of the lines that will divide each cell. linecolor : color, optional Color of the lines that will divide each cell. cbar : boolean, optional Whether to draw a colorbar. cbar_kws : dict of key, value mappings, optional Keyword arguments for `fig.colorbar`. cbar_ax : matplotlib Axes, optional Axes in which to draw the colorbar, otherwise take space from the main Axes. square : boolean, optional If True, set the Axes aspect to "equal" so each cell will be square-shaped. xticklabels, yticklabels : "auto", bool, list-like, or int, optional If True, plot the column names of the dataframe. If False, don't plot the column names. If list-like, plot these alternate labels as the xticklabels. If an integer, use the column names but plot only every n label. If "auto", try to densely plot non-overlapping labels. mask : boolean array or DataFrame, optional If passed, data will not be shown in cells where ``mask`` is True. Cells with missing values are automatically masked. ax : matplotlib Axes, optional Axes in which to draw the plot, otherwise use the currently-active Axes. kwargs : other keyword arguments All other keyword arguments are passed to ``ax.pcolormesh``. Returns ------- ax : matplotlib Axes Axes object with the heatmap. See also -------- clustermap : Plot a matrix using hierachical clustering to arrange the rows and columns. Examples -------- Plot a heatmap for a numpy array: .. plot:: :context: close-figs >>> import numpy as np; np.random.seed(0) >>> import seaborn as sns; sns.set() >>> uniform_data = np.random.rand(10, 12) >>> ax = sns.heatmap(uniform_data) Change the limits of the colormap: .. plot:: :context: close-figs >>> ax = sns.heatmap(uniform_data, vmin=0, vmax=1) Plot a heatmap for data centered on 0 with a diverging colormap: .. plot:: :context: close-figs >>> normal_data = np.random.randn(10, 12) >>> ax = sns.heatmap(normal_data, center=0) Plot a dataframe with meaningful row and column labels: .. plot:: :context: close-figs >>> flights = sns.load_dataset("flights") >>> flights = flights.pivot("month", "year", "passengers") >>> ax = sns.heatmap(flights) Annotate each cell with the numeric value using integer formatting: .. plot:: :context: close-figs >>> ax = sns.heatmap(flights, annot=True, fmt="d") Add lines between each cell: .. plot:: :context: close-figs >>> ax = sns.heatmap(flights, linewidths=.5) Use a different colormap: .. plot:: :context: close-figs >>> ax = sns.heatmap(flights, cmap="YlGnBu") Center the colormap at a specific value: .. plot:: :context: close-figs >>> ax = sns.heatmap(flights, center=flights.loc["January", 1955]) Plot every other column label and don't plot row labels: .. plot:: :context: close-figs >>> data = np.random.randn(50, 20) >>> ax = sns.heatmap(data, xticklabels=2, yticklabels=False) Don't draw a colorbar: .. plot:: :context: close-figs >>> ax = sns.heatmap(flights, cbar=False) Use different axes for the colorbar: .. plot:: :context: close-figs >>> grid_kws = {"height_ratios": (.9, .05), "hspace": .3} >>> f, (ax, cbar_ax) = plt.subplots(2, gridspec_kw=grid_kws) >>> ax = sns.heatmap(flights, ax=ax, ... cbar_ax=cbar_ax, ... cbar_kws={"orientation": "horizontal"}) Use a mask to plot only part of a matrix .. plot:: :context: close-figs >>> corr = np.corrcoef(np.random.randn(10, 200)) >>> mask = np.zeros_like(corr) >>> mask[np.triu_indices_from(mask)] = True >>> with sns.axes_style("white"): ... ax = sns.heatmap(corr, mask=mask, vmax=.3, square=True) t linewidthst edgecolortequalN(R4R(Rtgcat set_aspectR(R0RTRURVRWRXRJRLRMRt linecolorRNROtcbar_axtsquareRARDR1Rtkwargstplotter((s-lib/python2.7/site-packages/seaborn/matrix.pyR Ls     t_DendrogramPlottercB sYeZdZdZdZdZedZdZedZ dZ RS(sAObject for drawing tree of similarities between data rows/columnsc C s,||_|jdkr$|j}nt|tjrB|j}ntj|}tj|}||_||_ |j j |_ ||_ ||_ ||_||_ ||_|dkr|j|_n ||_|j|_dtj|j j dd} |j rt|j j} g|jD]} | | ^q'} |jrg|_| |_g|_| |_t|j j|_d|_q| |_g|_| |_g|_d|_t|j j|_n8gg|_|_gg|_|_d \|_|_|jd|_|jd|_ dS( sPlot a dendrogram of the relationships between the columns of data Parameters ---------- data : pandas.DataFrame Rectangular data ii iiR6tdcoordticoordN(R6R6(!RR>RRR RR)R8tarrayR0R+tmetrictmethodRtrotateR(tcalculated_linkagetlinkagetcalculate_dendrogramt dendrogramRuRRt reordered_indR@RCRARDRRFREtdependent_coordtindependent_coord( RSR0RRRRRRRR7t ticklabelsR((s-lib/python2.7/site-packages/seaborn/matrix.pyR]sR            !             cC sMtj|jdkr%tdntj|jd|jd|j}|S(Ni'sCThis will be slow... (gentle suggestion: "pip install fastcluster")RR( R)tproductR+t UserWarningRRRRR(RSR((s-lib/python2.7/site-packages/seaborn/matrix.pyt_calculate_linkage_scipyVs   cC sddl}d }|jdko-|j|k}|sE|jdkrg|j|jd|jd|jS|j|jd|jd|j}|SdS( Nitcentroidtmediantwardt euclideantsingleRR(RRR(t fastclusterRRtlinkage_vectorRR(RSRteuclidean_methodsRR((s-lib/python2.7/site-packages/seaborn/matrix.pyt_calculate_linkage_fastcluster^s     cC s-y|jSWntk r(|jSXdS(N(Rt ImportErrorR(RS((s-lib/python2.7/site-packages/seaborn/matrix.pyRos cC s tj|jdtdtj S(sCalculates a dendrogram based on the linkage matrix Made a separate function, not a property because don't want to recalculate the dendrogram every time it is accessed. Returns ------- dendrogram : dict Dendrogram dictionary as returned by scipy.cluster.hierarchy .dendrogram. The important key-value pairing is "reordered_ind" which indicates the re-ordering of the matrix tno_plottcolor_threshold(RRRR<R)tinf(RS((s-lib/python2.7/site-packages/seaborn/matrix.pyRvs cC s |jdS(s2Indices of the matrix, reordered by the dendrogramtleaves(R(RS((s-lib/python2.7/site-packages/seaborn/matrix.pyRsc C stdddd}|jrv|jdkrvtgt|j|jD]!\}}tt||^qF|}nFtgt|j|jD]!\}}tt||^q|}|j|t |j }t t t |j}|jrH|j jd|jd|d|jd|d|j|jn(|jd|d|jd|dtd |d td t|jd |jd |jd|jd|j|j|j}|j|jdd} t|jj dr|jj!|jj j"nt | dkrIt#| rIt$j%| ddnt |dkr}t#|r}t$j%|ddn|S(sPlots a dendrogram of the similarities between data on the axes Parameters ---------- ax : matplotlib.axes.Axes Axes object upon which the dendrogram is plotted Rg?Rtkitrighti g?RRRR@RCRERFRRRR(&R|RRRRvRRR"tadd_collectionR?RRaRRtset_ticks_positiontset_ylimtset_xlimt invert_xaxisRRR<RR@RCRERFRRARRDRRRRRRRR( RSRt line_kwargsRRtlinestnumber_of_leavestmax_dependent_coordRR((s-lib/python2.7/site-packages/seaborn/matrix.pyRsB  .  .     ( RRRR]RRtpropertyRRRR(((s-lib/python2.7/site-packages/seaborn/matrix.pyRs ?   Rtaveragec C s[t|d|d|d|d|d|d|}|dkrKtj}n|jd|S( s%Draw a tree diagram of relationships within a matrix Parameters ---------- data : pandas.DataFrame Rectangular data linkage : numpy.array, optional Linkage matrix axis : int, optional Which axis to use to calculate linkage. 0 is rows, 1 is columns. label : bool, optional If True, label the dendrogram at leaves with column or row names metric : str, optional Distance metric. Anything valid for scipy.spatial.distance.pdist method : str, optional Linkage method to use. Anything valid for scipy.cluster.hierarchy.linkage rotate : bool, optional When plotting the matrix, whether to rotate it 90 degrees counter-clockwise, so the leaves face right ax : matplotlib axis, optional Axis to plot on, otherwise uses current axis Returns ------- dendrogramplotter : _DendrogramPlotter A Dendrogram plotter object. Notes ----- Access the reordered dendrogram indices with dendrogramplotter.reordered_ind RRRRRRRN(RR(RRR( R0RRRRRRRR((s-lib/python2.7/site-packages/seaborn/matrix.pyRs $  t ClusterGridcB seZddddddddZdZdddZeddZeddZddZ edd Z d Z d Z d Z d ZdZRS(c  C st|tjr||_ntj||_|j|j||||_t|j||_|dkrd\} } | | f}nt j d||_ |j ||dd\|_ |_|j ||dd\|_|_|j|j d|dd} |j|jd|dd} |jdkr8dnd} |j dkrSdnd}tj| |dd d d d | d | |_|j j|j| ddd f|_|j j|jdd |df|_|jj|jjd|_d|_|j dk rF|j j|j| d|d f|_n|jdk r|j j|j| d |df|_n|j j|j| d|df|_|j j|jd|_d|_d|_dS(s=Grid object for organizing clustered heatmap input on to axesi tfigsizeRiiiitwspaceg{Gz?thspacet width_ratiost height_ratiosiN(i i (ii(RRR R0t format_datatdata2dR3R1R(RRtfigt_preprocess_colorst row_colorstrow_color_labelst col_colorstcol_color_labelst dim_ratiosRtGridSpectgst add_subplottax_row_dendrogramtax_col_dendrogramt set_axis_offt ax_row_colorst ax_col_colorst ax_heatmapRtdendrogram_rowtdendrogram_col(RSR0t pivot_kwstz_scoretstandard_scaleRR R R1RRRRtnrowstncols((s-lib/python2.7/site-packages/seaborn/matrix.pyR]sN    $$   ,,     $ $* cC sd}|dk rt|tjtjfr|dkrL|j|j}n|j|j}|jd}t|tjrt |j}|j j }q|j dkrdg}n |j g}|j }nt |}n||fS(sAPreprocess {row/col}_colors to extract labels and convert colors.iRR6N(R(RRR R!tlocRR&tfillnaR"R>RRR%(RSR0RRR((s-lib/python2.7/site-packages/seaborn/matrix.pyR /s      cC s|dk r|j|}n|}|dk rK|dk rKtdn|dk rl|j||}n|dk r|j||}n|S(s,Extract variables from data or use directly.s:Cannot perform both z-scoring and standard-scaling on dataN(R(tpivotR#RR(RSR0RRRR((s-lib/python2.7/site-packages/seaborn/matrix.pyRNs    icC sS|dkr|}n |j}||j|j}|dkrH|S|jSdS(sStandarize the mean and variance of the data axis Parameters ---------- data2d : pandas.DataFrame Data to normalize axis : int Which axis to normalize across. If 0, normalize across rows, if 1, normalize across columns. Returns ------- normalized : pandas.DataFrame Noramlized data with a mean of 0 and variance of 1 across the specified axis. iN(R>tmeantstd(RRtz_scored((s-lib/python2.7/site-packages/seaborn/matrix.pyRbs    cC sc|dkr|}n |j}|j}|||j|j}|dkrX|S|jSdS(sVDivide the data by the difference between the max and min Parameters ---------- data2d : pandas.DataFrame Data to normalize axis : int Which axis to normalize across. If 0, normalize across rows, if 1, normalize across columns. vmin : int If 0, then subtract the minimum of the data before dividing by the range. Returns ------- standardized : pandas.DataFrame Noramlized data with a mean of 0 and variance of 1 across the specified axis. iN(R>R`Ra(RRt standardizedtsubtract((s-lib/python2.7/site-packages/seaborn/matrix.pyRs     g?c C s||}td|d}d|}d|}|dkrL||g} n ||g} |dk rt| |g7} n| dg7} | S(s@Get the proportions of the figure taken up by each axes g@g?g?iN(R`R(( RSt side_colorsRRtside_colors_ratiotfigdimRtcolorbar_widthtcolorbar_heighttratios((s-lib/python2.7/site-packages/seaborn/matrix.pyRs       ic C s%td|DrJttj|}t|}t|d}n't|}d}t|}|g}tdt|D}tjg|D]}|D]}||^qq} ||f} | j | } | dd|f} |dkr | j } nt j j |} | | fS(sTurns a list of colors into a numpy matrix and matplotlib colormap These arguments can now be plotted using heatmap(matrix, cmap) and the provided colors will be plotted. Parameters ---------- colors : list of matplotlib colors Colors to label the rows or columns of a dataframe. ind : list of ints Ordering of the rows or columns, to reorder the original colors by the clustered dendrogram order axis : int Which axis this is labeling Returns ------- matrix : numpy.array A numpy array of integer values, where each corresponds to a color from the originally provided list of colors cmap : matplotlib.colors.ListedColormap cs s$|]}tt|tVqdS(N(t issubclassttypeR"(RR((s-lib/python2.7/site-packages/seaborn/matrix.pys siics s!|]\}}||fVqdS(N((RRR((s-lib/python2.7/site-packages/seaborn/matrix.pys sN(tanyRt itertoolstchainR?R|t enumerateR)RtreshapeR>RRRe( RtindRt all_colorsRRtcolor_to_valueRptctmatrixR+RV((s-lib/python2.7/site-packages/seaborn/matrix.pytcolor_list_to_matrix_and_cmaps$     *   cO s0d|krd|d>> import seaborn as sns; sns.set(color_codes=True) >>> iris = sns.load_dataset("iris") >>> species = iris.pop("species") >>> g = sns.clustermap(iris) Use a different similarity metric: .. plot:: :context: close-figs >>> g = sns.clustermap(iris, metric="correlation") Use a different clustering method: .. plot:: :context: close-figs >>> g = sns.clustermap(iris, method="single") Use a different colormap and ignore outliers in colormap limits: .. plot:: :context: close-figs >>> g = sns.clustermap(iris, cmap="mako", robust=True) Change the size of the figure: .. plot:: :context: close-figs >>> g = sns.clustermap(iris, figsize=(6, 7)) Plot one of the axes in its original organization: .. plot:: :context: close-figs >>> g = sns.clustermap(iris, col_cluster=False) Add colored labels: .. plot:: :context: close-figs >>> lut = dict(zip(species.unique(), "rbg")) >>> row_colors = species.map(lut) >>> g = sns.clustermap(iris, row_colors=row_colors) Standardize the data within the columns: .. plot:: :context: close-figs >>> g = sns.clustermap(iris, standard_scale=1) Normalize the data within the rows: .. plot:: :context: close-figs >>> g = sns.clustermap(iris, z_score=0) RRR R RRR1RRRSRARBRCRD(RR(R0RRRRRRRORARBRCRDR R R1RR((s-lib/python2.7/site-packages/seaborn/matrix.pyR ws     (,Rt __future__RR1t matplotlibRtmatplotlib.collectionsRtmatplotlib.pyplottpyplotRRtnumpyR)tpandasRt scipy.clusterRR6RtaxisgridRtutilsRRRR t external.sixR t__all__RRR%R3tobjectR4R(R=R<R RRRR (((s-lib/python2.7/site-packages/seaborn/matrix.pytsF    "     %    +