使用稀疏的数据添加相关系数

Question

我试图将一个配对式的元素组合在一起，以证明各组值之间的相关性。但是，我遇到了计算相关系数的问题，因为数据很少，只有一些组对于所有值都有条目（请参见下面的示例DF）。

group,PMTS049_X1,PMTS049_X2,PMTS049_X3,PMTS050_X1,PMTS050_X2,PMTS050_X3,PMTS051_X1,PMTS051_X2,PMTS051_X3
A-1155463_0.0045724735 uM,,,,,,,1.04971460376213,1.179179562152282,1.108186227034926
A-1155463_0.0137174204 uM,,,,,,,0.8988914257248843,1.1081975403870097,1.1212163392527827
A-1155463_0.0411522626 uM,,,,,,,0.7568287138412346,1.3430358264779123,0.6976356991070309
A-1155463_0.1234567894 uM,,,,,,,0.8722751660077679,1.2813232679016255,0.7168068340749701
A-1155463_0.3703703698 uM,,,,,,,1.0048500564303646,0.9822764937914188,0.7848035710436727
A-1155463_1.1111111109 uM,,,,,,,0.714489988829366,1.0212592818965238,0.9073404868051366
A-1155463_10.0 uM,,,,,,,0.8653477680327408,1.0148035026211837,1.078068956627115
A-1155463_3.3333333328 uM,,,,,,,0.7886497200478821,1.2509355686549108,1.102089542546691
AR00536370_0.004572474 uM,0.904860165175536,0.630159462892308,0.9489118135217488,,,,,,
AR00536370_0.013717421 uM,1.0775483229823892,1.0225981211028854,0.882809722682908,,,,,,
AR00536370_0.041152263 uM,0.651388505631196,0.6464619116451127,0.9482309449208448,,,,,,
AR00536370_0.12345679 uM,1.1258840716614356,1.086943922239876,0.9390622801310888,,,,,,
AR00536370_0.37037037 uM,0.8376593712267018,0.6137443429066174,1.1138346723430423,,,,,,
AR00536370_1.111111111 uM,0.889524913610136,0.614134774405636,1.024540177391546,,,,,,
AR00536370_10.0 uM,1.1265016447481484,1.0079550835923723,1.020468601391634,,,,,,
AR00536370_3.333333333 uM,0.7712691128712621,0.5893153808880822,0.9392664499480852,,,,,,
AZ-628_0.0045724738 uM,0.9076622425465212,1.0934450174529409,0.8998208357308275,0.5434529688654449,1.0562636134080905,0.7715563402897756,0.8276719202512567,1.186335347522918,0.8472096021168107
AZ-628_0.0137174214 uM,1.0963705071916314,0.7842348321294206,0.8315757098390226,1.1030070502236278,0.9908839527491846,0.6890801893628911,1.0576768873554188,0.6458275700418388,0.6252105837953932
AZ-628_0.0411522642 uM,0.4638659293322327,0.448229525781371,0.7834117035243882,0.6719774212414784,0.7929078280555364,0.5459489088307414,0.5319067062568421,0.7597553877201356,0.7555114458518108
AZ-628_0.1234567905 uM,0.7477873760694489,0.6876014256016453,0.7062173803657827,0.638953551272889,0.7319274702673808,0.5026056304516787,0.670610155828824,0.7502067504385757,0.728226016276163
AZ-628_0.3703703714 uM,0.6866247604514487,0.6283063570534719,0.4496522225873166,0.6832591578588455,0.5954091111794844,0.3100417445307057,0.6715651569282897,0.618884745101185,0.3682229011490131
AZ-628_1.1111111121 uM,0.5067878800206379,0.4427746455438304,0.3437829115127019,0.5443902391591303,0.4502681586077313,0.4061182601803592,0.5963543157154589,0.54950245634917,0.2611084549558647

# copy the data to the clipboard and use
df = pd.read_clipboard(sep=',')

到目前为止，我想到的是以下内容：

def corr_func(x, y, **kws):
    r, _ = stats.pearsonr(x, y)
    ax = plt.gca()
    ax.annotate("{:.2f}".format(r),
            xy=(.3, .45), xycoords=ax.transAxes, size=30)

def plot_corrplot(df):
    g = sns.PairGrid(df, diag_sharey=False, corner=False)
    g.map_lower(sns.scatterplot, s=10, color='black')
    g.map_diag(sns.kdeplot, color='grey')
    g.map_upper(corr_func)
    plt.show()

散点图和KDE图显示，但是，由于NAN的存在，我无法正确计算相关性：

ValueError: array must not contain infs or NaNs

本质上，我需要做的是计算该系数在忽略任何未打开的条目（NAN）的同时。

下面显示了一个示例图，在不映射corr_function的情况下生成。在这种情况下，皮尔逊系数应在上图中。