2 Seaborn

该库对Matplotlib进行了更高级的封装,使得用户可以在少量代码下绘制出复杂的图形。

2.1 图类型

Seaborn提供的可绘制图类型包括统计关系型、数据分布型、分类数据型、回归模型分析型和多子图网格型。

2.1.1 关系型
2.1.1.1 scatterplot()

seaborn.scatterplot(data=None, *, x=None, y=None, hue=None, size=None, style=None, palette=None, hue_order=None, hue_norm=None, sizes=None, size_order=None, size_norm=None, markers=True, style_order=None, legend='auto', ax=None, **kwargs)

  • data 绘图源数据,可以为pandas.DataFrame,numpy.ndarray等序列或映射
  • x、y 确定两坐标轴对应的变量
  • hue 指定某一变量,其不同值对应图中的不同颜色
  • size 指定某一变量,其不同值对应图中不同点的大小
  • style 指定某一变量,其不同值对应图中的不同模式(如实线与虚线)
  • palette 指定将hue中元素到不同颜色之间的映射
  • sizes 确定不同size选用的方案
  • markers 确定不同点类型的选用方案
  • legend "auto","brief", "full", 或False
  • **kwargs 其他传给matplotlib.pyplot.scatter()函数的关键字参数

以下是官网的使用例:

import seaborn as sns# 一些方便的设置def plot(plot_func, fig_size, **kwargs):    plt.figure(figsize=fig_size, dpi=200)    ax = plt.subplot(111)    plot_func(ax=ax, **kwargs)    plt.show()# 加载数据tips = sns.load_dataset("tips")# 这个数据集一共有七列,即# | total_bill | tip | sex | smoker | day | time | size |# 不指定其他参数plot(sns.scatterplot,      (10, 10),      data=tips, x="total_bill", y="tip")

image-20230825122707730

# 指定hue、style、sizeplot(sns.scatterplot,      (10, 10),      data=tips, x="total_bill", y="tip", hue="time",     style="time", size="size")

image-20230825123005219

# 指定hue为数值(numerical)而非分类(categorical)类型plot(sns.scatterplot,      (10, 10),      data=tips,      x="total_bill", y="tip",     hue=tips.eval("tip / total_bill").rename("tip_rate")     )

image-20230825123258728

2.1.1.2 relplot()

seaborn.relplot(data=None, *, x=None, y=None, hue=None, size=None, style=None, units=None, row=None, col=None, col_wrap=None, row_order=None, col_order=None, palette=None, hue_order=None, hue_norm=None, sizes=None, size_order=None, size_norm=None, markers=None, dashes=None, style_order=None, legend='auto', kind='scatter', height=5, aspect=1, facet_kws=None, **kwargs)

  • units 分类变量的指标,一般用于绘制多次重复实验的结果
  • row、col 用于定义不同行和列的对应指标
  • dashes 定义线型
  • kind 选择线图或散点图
  • height 每个子图的高(inch)
  • aspect 每个子图的宽高比,width = aspect * height
  • facet_kw 传入FacetGrid的关键字参数
# 指定hue,并分别指定行和列sns.relplot(data=tips,             x="total_bill", y="tip",             hue="day", col="time", row="sex"            )

image-20230825130419372

# 指定hue、col和rowsns.relplot(data=tips,             x="total_bill",             y="tip",             col="time",            hue="time",             size="size",             style="sex",            palette=["b", "r"],             sizes=(10, 100)            )

image-20230825131003557

2.1.1.3 lineplot()

seaborn.lineplot(data=None, *, x=None, y=None, hue=None, size=None, style=None, units=None, palette=None, hue_order=None, hue_norm=None, sizes=None, size_order=None, size_norm=None, dashes=True, markers=None, style_order=None, estimator='mean', errorbar =('ci', 95), n_boot=1000, seed=None, orient='x', sort=True, err_style='band', err_kws=None, legend='auto', ci='deprecated', ax=None, **kwargs)

  • estimator 估计方法,类型为callable或None,用于估计图线在某一处的值
  • errorbar 选择对应的errorbar方法,或传入一个元组
  • n_boot 计算置信区间时所使用的自助次数
  • seed 自助法所使用的随机种子
  • orient 目标函数的独立变量
  • sort 对x或y变量进行排序
flights = sns.load("flights")flights.head()
year month passengers
0 1949 Jan 112
1 1949 Feb 118
2 1949 Mar 132
3 1949 Apr 129
4 1949 May 121
may_flights = floghts.query("month == 'May'")may_flights.head()
year month passengers
4 1949 May 121
16 1950 May 125
28 1951 May 172
40 1952 May 183
52 1953 May 229
plt.figure(dpi=300, figsize=(18, 9))ax = plt.subplot(121)sns.lineplot(data=may_flights, x="year", y="passengers", ax=ax)ax = plt.subplot(122)sns.lineplot(data=flights, x="year", y="passengers", ax=ax)plt.tight_layout()

image-20230827112928236

flights_wide = flights.pivot("year", "month", "passengers")flights_wide.head()
month Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
year
1949 112 118 132 129 121 135 148 148 136 119 104 118
1950 115 126 141 135 125 149 170 170 158 133 114 140
1951 145 150 178 163 172 178 199 199 184 162 146 166
1952 171 180 193 181 183 218 230 242 209 191 172 194
1953 196 196 236 235 229 243 264 272 237 211 180 201
plt.figure(dpi=300, figsize=(12, 12))sns.lineplot(data=flights_wide)plt.grid()

image-20230827113329755

fmri = sns.load_dataset("fmri")plt.figure(dpi=300, figsize=(16, 16))plt.subplot(2, 2, 1)sns.lineplot(data=fmri, x="timepoint", y="signal", hue="event")plt.subplot(2, 2, 2)sns.lineplot(data=fmri, x="timepoint", y="signal", hue="region", style="event")plt.subplot(2, 2, 3)sns.lineplot(data=fmri, x="timepoint", y="signal", hue="region", style="event", markers=True, dashes=False)plt.subplot(2, 2, 4)sns.lineplot(data=fmri, x="timepoint", y="signal", hue="region", style="event", markers=True, err_style="bars")

image-20230827114411080

值得注意的是,笔者使用的seaborn版本和所查文档的版本不同,lineplot函数中没有errorbar参数,而是error_style。

import matplotlib as mpldots = sns.load_dataset("dots").query("align == 'dots'")plt.figure(dpi=300, figsize=(16, 16))plt.subplot(2, 2, 1)sns.lineplot(data=dots, x="time", y="firing_rate", hue="coherence", style="choice")plt.subplot(2, 2, 2)sns.lineplot(data=dots.query("coherence > 0"),              x="time", y="firing_rate",              hue="coherence", style="choice",             palette="flare", hue_norm=mpl.colors.LogNorm())plt.subplot(2, 2, 3)palette = sns.color_palette("mako_r", 6)sns.lineplot(data=dots,              x="time", y="firing_rate",              hue="coherence", style="choice",             palette=palette)plt.subplot(2, 2, 4)sns.lineplot(data=dots,              x="time", y="firing_rate",              size="coherence", hue="choice",             legend="full")

image-20230827115241283

在默认情况下,seaborn会对输入的x变量进行排序,在一些特定情况下,我们并不希望进行这样的操作。下面就是一个例子

from tqdm import tqdmITER_NUM = 2000x, y = np.random.normal(size=(2, ITER_NUM)).cumsum(axis=1)plt.figure(dpi=300, figsize=(16, 8))# 直接绘制ax = plt.subplot(121)sns.lineplot(x=x, y=y, sort=False, lw=1, ax=ax)# 添加渐变颜色以显示轨迹的新旧ax = plt.subplot(122)hue = np.linspace(0, 5, ITER_NUM) / 5 + 1e-5for i in tqdm(range(0, ITER_NUM)):    sns.lineplot(x=x[i:i+2], y=y[i:i+2],                  sort=False,                  lw=1,                  color=plt.get_cmap("BuPu")(hue[i]),                  legend=False,                  ax=ax)

image-20230827123256923

2.1.2 数据分布型
2.1.2.1 displot()
def displot(        data=None, *,        # Vector variables        x=None, y=None, hue=None, row=None, col=None, weights=None,        # Other plot parameters        kind="hist", rug=False, rug_kws=None, log_scale=None, legend=True,        # Hue-mapping parameters        palette=None, hue_order=None, hue_norm=None, color=None,        # Faceting parameters        col_wrap=None, row_order=None, col_order=None,        height=5, aspect=1, facet_kws=None,        **kwargs,)

绘制分布图的函数。

penguins = sns.load_dataset("penguins")sns.displot(data=penguins, x="flipper_length_mm", kde=True, facet_kws={"size":8})plt.xlabel("flipper_length_mm")plt.title("Histogram and KDE curve")

image-20230827125953265

sns.displot(data=penguins, x="flipper_length_mm", kind="ecdf", facet_kws={"size":8})plt.xlabel("flipper_length_mm")plt.title("ECDF curve")

image-20230827163521536

sns.displot(data=penguins, x="flipper_length_mm", y="bill_length_mm", bins=20           , height=8)plt.title("Heat Map")

image-20230827163714881

sns.displot(data=penguins, x="flipper_length_mm", y="bill_length_mm", height=8, kind="kde", rug=True)plt.title("KDE Contour Map")

image-20230827163930646

2.1.2.2 histpot()
def histplot(        data=None, *,        # Vector variables        x=None, y=None, hue=None, weights=None,        # Histogram computation parameters        stat="count", bins="auto", binwidth=None, binrange=None,        discrete=None, cumulative=False, common_bins=True, common_norm=True,        # Histogram appearance parameters        multiple="layer", element="bars", fill=True, shrink=1,        # Histogram smoothing with a kernel density estimate        kde=False, kde_kws=None, line_kws=None,        # Bivariate histogram parameters        thresh=0, pthresh=None, pmax=None, cbar=False, cbar_ax=None, cbar_kws=None,        # Hue mapping parameters        palette=None, hue_order=None, hue_norm=None, color=None,        # Axes information        log_scale=None, legend=True, ax=None,        # Other appearance keywords        **kwargs,) -> matplotlib.axes.Axes: ...
penguins = penguins = sns.load_dataset("penguins")plt.figure(figsize=(16, 16), dpi=300)plt.subplot(2, 2, 1)# 还可以使用 binwidth 间接控制 bins 的数量sns.histplot(data=penguins, x="flipper_length_mm", bins=30, kde=True)plt.subplot(2, 2, 2)# 使用 hue 区分不同的分布,并可以指定 multiple 为 "stack" 避免遮挡的情况sns.histplot(data=penguins, x="flipper_length_mm", hue="species", multiple="stack")plt.subplot(2, 2, 3)# 指定 common_norm 将所有数据进行归一化sns.histplot(    penguins, x="bill_length_mm", hue="island", element="step",    stat="density", common_norm=False,)plt.subplot(2, 2, 4)# 在离散的数据下绘制条形图sns.histplot(data=penguins, x="island", hue="sex", multiple="dodge", shrink=.8)

image-20230827174536442

2.1.2.3 kdeplot()
def kdeplot(        x=None,  # Allow positional x, because behavior will not change with reorg        *,        y=None,        shade=None,  # Note "soft" deprecation, explained below        vertical=False,  # Deprecated        kernel=None,  # Deprecated        bw=None,  # Deprecated        gridsize=200,  # TODO maybe depend on uni/bivariate?        cut=3, clip=None, legend=True, cumulative=False,        shade_lowest=None,  # Deprecated, controlled with levels now        cbar=False, cbar_ax=None, cbar_kws=None,        ax=None,        # New params        weights=None,  # TODO note that weights is grouped with semantics        hue=None, palette=None, hue_order=None, hue_norm=None,        multiple="layer", common_norm=True, common_grid=False,        levels=10, thresh=.05,        bw_method="scott", bw_adjust=1, log_scale=None,        color=None, fill=None,        # Renamed params        data=None, data2=None,        # New in v0.12        warn_singular=True,        **kwargs,) -> matplotlib.axes.Axes: ...
iris = sns.load_dataset("iris")tips = sns.load_dataset("tips")geyser = sns.load_dataset("geyser")plt.figure(dpi=300, figsize=(24, 16))plt.subplot(2, 3, 1)# 绘制多个分布密度估计曲线并调整平滑带宽sns.kdeplot(data=iris, bw_adjust=.2)plt.subplot(2, 3, 2)# 堆叠多个分布估计曲线sns.kdeplot(data=tips, x="total_bill", hue="time", multiple="stack")plt.subplot(2, 3, 3)# 归一化多个分布估计曲线sns.kdeplot(data=tips, x="total_bill", hue="time", multiple="fill")plt.subplot(2, 3, 4)# 绘制多个累积分布估计曲线sns.kdeplot(    data=tips, x="total_bill", hue="time",    cumulative=True, common_norm=False, common_grid=True)plt.subplot(2, 3, 5)# 使用数值类型的 huesns.kdeplot(   data=tips, x="total_bill", hue="size",   fill=True, common_norm=False, palette="crest",   alpha=.5, linewidth=0,)plt.subplot(2, 3, 6)# 绘制多个二维分布的等高线图,并设置参数sns.kdeplot(    data=geyser, x="waiting", y="duration", hue="kind",    levels=10, thresh=.1,)

image-20230827175425396

2.1.2.4 ecdfplot()
def ecdfplot(        data=None, *,        # Vector variables        x=None, y=None, hue=None, weights=None,        # Computation parameters        stat="proportion", complementary=False,        # Hue mapping parameters        palette=None, hue_order=None, hue_norm=None,        # Axes information        log_scale=None, legend=True, ax=None,        # Other appearance keywords        **kwargs,) -> matplotlib.axes.Axes: ...

上文中已有使用cedfplot绘制的例子,在此不再赘述。

2.1.2.5 rugplot()
def rugplot(    x=None,  # Allow positional x, because behavior won't change    *,    height=.025, axis=None, ax=None,    # New parameters    data=None, y=None, hue=None,    palette=None, hue_order=None, hue_norm=None,    expand_margins=True,    legend=True,  # TODO or maybe default to False?    # Renamed parameter    a=None,    **kwargs) -> matplotlib.axes.Axes: ...
plt.figure(dpi=300, figsize=(16, 8))plt.subplot(1, 2, 1)# rugplot 往往是在scatterplot的基础之上绘制的sns.scatterplot(data=tips, x="total_bill", y="tip", hue="time")sns.rugplot(data=tips, x="total_bill", y="tip", hue="time", height=0.05)plt.subplot(1, 2, 2)# 在面对大数据集时,需要适当的将线条粗细降低,将线条透明度调高diamonds = sns.load_dataset("diamonds")sns.scatterplot(data=diamonds, x="carat", y="price", s=5)sns.rugplot(data=diamonds, x="carat", y="price", lw=1, alpha=.005)

image-20230827180211885

2.1.3 分类数据型
2.1.3.1 stripplot()
def stripplot(    *,    x=None, y=None,    hue=None, data=None,    order=None, hue_order=None,    jitter=True, dodge=False, orient=None, color=None, palette=None,    size=5, edgecolor="gray", linewidth=0, ax=None,    **kwargs): ...
plt.figure(dpi=300, figsize=(16, 8))# 添加类别plt.subplot(1, 2, 1)sns.stripplot(data=tips, x="total_bill", y="day", hue="sex")# 设置 dodge 为 True 以做到不同 hue 的分离,调整 jitter 为 False 使得数据点纵向不堆叠plt.subplot(1, 2, 2)sns.stripplot(data=tips, x="total_bill", y="day", hue="sex", dodge=True, jitter=False             , marker='D', alpha=.1, s = 10)

image-20230827183617984

2.1.3.2 swarmplot()
def swarmplot(    *,    x=None, y=None,    hue=None, data=None,    order=None, hue_order=None,    dodge=False, orient=None, color=None, palette=None,    size=5, edgecolor="gray", linewidth=0, ax=None,    **kwargs): ...
2.1.3.3 boxplot()
def boxplot(    *,    x=None, y=None,    hue=None, data=None,    order=None, hue_order=None,    orient=None, color=None, palette=None, saturation=.75,    width=.8, dodge=True, fliersize=5, linewidth=None,    whis=1.5, ax=None,    **kwargs): ...

箱线图可以显示数据的主要统计量,如中位数、四分位数、离群值等,如下图所示。

img

df = sns.load_dataset("titanic")plt.figure(dpi=300, figsize=(16, 8))plt.subplot(1, 2, 1)sns.boxplot(data=df, x="fare", y="deck", hue="deck", dodge=False)plt.subplot(1, 2, 2)sns.boxplot(    data=df, x="age", y="class",    notch=True, 								# 将箱线图类型设置为“葫芦”的形状    showcaps=False,								# 是否显示头尾两条垂直线    flierprops={"marker": "x"},					# 设置离群值点属性    boxprops={"facecolor": (.4, .6, .8, .5)},	# 设置颜色,分别为 RGB 和透明度    medianprops={"color": "coral"},				# 设置中位数属性)

image-20230827184512858

2.1.3.4 violinplot()
>def violinplot(   *,   x=None, y=None,   hue=None, data=None,   order=None, hue_order=None,   bw="scott", cut=2, scale="area", scale_hue=True, gridsize=100,   width=.8, inner="box", split=False, dodge=True, orient=None,   linewidth=None, color=None, palette=None, saturation=.75,   ax=None, **kwargs,>): ...

提琴图除了拥有箱线图的优点以外还可以很好的反应数据的分布特征

df = sns.load_dataset("titanic")plt.figure(dpi=300, figsize=(16, 16))plt.subplot(2, 2, 1)sns.violinplot(data=df, x="deck", y="age", hue="alive", split=True)# 画密度线plt.subplot(2, 2, 2)sns.violinplot(data=df, x="deck", y="age", hue="alive", split=True, inner="stick")# 四分位数plt.subplot(2, 2, 3)sns.violinplot(data=df, x="deck", y="age", hue="alive", split=True, inner="quartile")# 画独立实验点 (Individual Observartions)plt.subplot(2, 2, 4)sns.violinplot(data=df, x="deck", y="age", hue="alive", split=True, inner="point")

image-20230827185201016

p.s. 这玩意儿感觉有点像涡虫(下图,还怪可爱的)

image-20230827185753928

2.1.3.5 boxenplot()
def boxenplot(       *,       x=None, y=None,       hue=None, data=None,       order=None, hue_order=None,       orient=None, color=None, palette=None, saturation=.75,       width=.8, dodge=True, k_depth='tukey', linewidth=None,       scale='exponential', outlier_prop=0.007, trust_alpha=0.05, showfliers=True,       ax=None, **kwargs): ...
df = sns.load_dataset("diamonds")plt.figure(dpi=300, figsize=(16, 8))# 调整每个箱的宽度plt.subplot(1, 2, 1)sns.boxenplot(data=df, x="carat", y="cut", scale="linear")# 调整箱的个数plt.subplot(1, 2, 2)sns.boxenplot(data=df, x="carat", y="cut", k_depth="trustworthy")

image-20230827190934393

2.1.3.6 pointplot()
def pointplot(    data=None, *, x=None, y=None, hue=None, order=None, hue_order=None,    estimator="mean", errorbar=("ci", 95), n_boot=1000, units=None, seed=None,    markers="o", linestyles="-", dodge=False, join=True, scale=1,    orient=None, color=None, palette=None, errwidth=None, ci="deprecated",    capsize=None, label=None, ax=None,):...
df = sns.load_dataset("penguins")plt.figure(dpi=300, figsize=(16, 16))plt.subplot(2, 2, 1)sns.pointplot(data=df, x="island", y="body_mass_g")plt.subplot(2, 2, 2)sns.pointplot(data=df, x="island", y="body_mass_g", hue="sex")# 使用标准差绘制误差线plt.subplot(2, 2, 3)sns.pointplot(data=df, x="sex", y="body_mass_g", hue="island", dodge=True, errorbar="sd")# 自定义一些图形参数plt.subplot(2, 2, 4)sns.pointplot(data=df, x="body_mass_g", y="island", errorbar=("pi", 100),              capsize=.4,               join=False,              color=".5"              )

image-20230829164833094

2.1.3.7 barplot()
def barplot(    data=None, *, x=None, y=None, hue=None, order=None, hue_order=None,    estimator="mean", errorbar=("ci", 95), n_boot=1000, units=None, seed=None,    orient=None, color=None, palette=None, saturation=.75, width=.8,    errcolor=".26", errwidth=None, capsize=None, dodge=True, ci="deprecated",    ax=None,    **kwargs,):...
df = sns.load_dataset("penguins")plt.figure(dpi=300, figsize=(16, 16))# 其使用方法和 pointplot 类似plt.subplot(2, 2, 1)sns.barplot(data=df, x="island", y="body_mass_g")plt.subplot(2, 2, 2)sns.barplot(data=df, x="island", y="body_mass_g", hue="sex")plt.subplot(2, 2, 3)sns.barplot(data=df, x="sex", y="body_mass_g", hue="island", dodge=True, errorbar="sd")plt.subplot(2, 2, 4)sns.barplot(data=df, x="body_mass_g", y="island", errorbar=("pi", 100),              capsize=.4,               edgecolor=".5",              errcolor=".5",              facecolor=(0, 0, 0, 0)              )

image-20230829165215794

2.1.3.8 countplot()
def countplot(    data=None, *, x=None, y=None, hue=None, order=None, hue_order=None,    orient=None, color=None, palette=None, saturation=.75, width=.8,    dodge=True, ax=None, **kwargs):...

相比之下 countplot() 的使用就更加简单:

sns.set_theme(style="whitegrid")df = sns.load_dataset("titanic")plt.figure(dpi=300, figsize=(16, 8))plt.subplot(2, 2, 1)sns.countplot(data=df, x="class", hue="alive")plt.subplot(2, 2, 2)sns.countplot(data=df, y="deck", hue="alive")

image-20230829165539995

2.1.4 回归模型分析型
2.1.4.1 lmplot()

该函数在绘制数据散点的同时对其调用回归模型对其进行拟合,它结合了regplot()和FacetGrid(),从其参数列表也可以看出。

penguins = sns.load_dataset("penguins")sns.lmplot(data=penguins, x="bill_length_mm", y="bill_depth_mm", hue="species")

image-20230829170126762

可以绘制多子图同时观察不同类别下的数据拟合情况

sns.lmplot(data=penguins,            x="bill_length_mm",           y="bill_depth_mm",           hue="species",           col="sex",           height=8)

image-20230829170327203

2.1.4.2 regplot()

lmplot()以涵盖其全部功能,故不再赘述

2.1.4.3 residplot()

residplot()绘制你和结果的残差。

mpg = sns.load_dataset("mpg")plt.figure(dpi=300, figsize=(16, 8))plt.subplot(1, 2, 1)plt.title("regplot()")sns.regplot(data=mpg, x="weight", y="displacement")plt.subplot(1, 2, 2)plt.title("residplot()")sns.residplot(data=mpg,               x="weight",               y="displacement",              order=2)

image-20230829171407772

2.2 多子图网格型图

2.2.1 FacetGrid 类
Methods or Attributes Type Functionality
__init__ Method 初始化matplotlib的figure以及FaceGrid对象
add_lengnd Method 绘制图例
apply Method 将grid传给某个用户自定义函数,然后返回自身
dspine Method 去除坐标轴
facet_axis Method 返回某一个axis
facet_data Method 返回含有所有facet的数据和具名indices的生成器
map Method 调用某一个绘图函数作用于facet中的数据的一个子集
map_dataframe Method 与.map类似
pipe Method 将grid传入一个用户自定义函数并返回对应的值
refline Method 在每个子图中添加参考线
savefig Method 存储绘图结果
set Method 设置每个子图Axes的参数
set_axis_labels Method 设置子图的两轴标签
set_titles Method 设置子图标题
set_xlabels / set_ylabels Method 设置子图的x轴或y轴标签
set_xticklabels / set_yticklabels Method 设置子图的x轴或y轴刻度标签
tick_params Method 设置刻度参数
tight_layout Method 调整子图的布局
ax Attribute 未设置facet变量时的Axes
axes Attribute Axes数组
axes_dict Attribute 包含Axes的字典
figure Attribute 可通过此调整grid背后的mpl.figure.Figure对象
legend Attribute 可通过此调整mpl.legend.Legend对象

__init__()

def __init__(        self, data, *,        row=None, col=None, hue=None, col_wrap=None,        sharex=True, sharey=True, height=3, aspect=1, palette=None,        row_order=None, col_order=None, hue_order=None, hue_kws=None,        dropna=False, legend_out=True, despine=True,        margin_titles=False, xlim=None, ylim=None, subplot_kws=None,        gridspec_kws=None,    ):...

示例

tips = sns.load_dataset("tips")def annotate(data, **kws):    n = len(data)    ax = plt.gca()    ax.text(.1, .6, f"N = {n}", transform=ax.transAxes)g = sns.FacetGrid(data=tips, col="time", row="sex", margin_titles=True)g.set_titles(col_template="{col_name}", row_template="{row_name}")g.map_dataframe(sns.scatterplot, x="total_bill", y="tip", hue="sex")g.map_dataframe(annotate)g.add_legend()g.refline(y=tips["tip"].median())

image-20230829174634490

2.2.2 PairGrid 类
Methods or Attributes Type Functionality
__init__ Method 初始化matplotlib的figure以及FaceGrid对象
… … …
map_diag Method 将绘图函数作用于对角线的子图
map_lower Method 将函数作用于下三角部分的子图(不包括对角线)
map_upper Method 将函数作用于上三角部分的子图(不包括对角线)
map_offdiag Method 将函数作用域除对角线以外的所有子图
… … …

__init__

def __init__(        self, data, *, hue=None, vars=None, x_vars=None, y_vars=None,        hue_order=None, palette=None, hue_kws=None, corner=False, diag_sharey=True,        height=2.5, aspect=1, layout_pad=.5, despine=True, dropna=False,    ):...

一些示例

g = sns.PairGrid(penguins, diag_sharey=False)g.map_diag(sns.kdeplot)g.map_lower(sns.kdeplot)g.map_upper(sns.scatterplot)

image-20230829175350442

# 仅仅绘制下三角部分g = sns.PairGrid(penguins, diag_sharey=False, corner=True)g.map_diag(sns.kdeplot)g.map_lower(sns.kdeplot)

image-20230829175514974

# 使多个函数作用在某些子图中g = sns.PairGrid(penguins, hue="species")g.map_diag(sns.histplot)g.map_offdiag(sns.scatterplot, size=penguins["sex"])g.add_legend(title="", adjust_subtitles=True)

image-20230829175625563

# 设置不同子图的 hue# 并使用 variables 控制传入的变量x_vars = ["body_mass_g", "bill_length_mm", "flipper_length_mm"]y_vars = ["body_mass_g", "bill_length_mm"]g = sns.PairGrid(penguins, hue="body_mass_g", x_vars=x_vars, y_vars=y_vars)g.map_diag(sns.histplot, hue=None, color=".3")g.map_offdiag(sns.scatterplot)g.add_legend()

image-20230829180023019

2.3 绘图风格、颜色主题和绘图元素缩放比例

2.3.1 绘图风格
2.3.1.1 seaborn.set_theme()
def set_theme(context="notebook",		# string or dict; Scaling parameters              style="darkgrid", 		# string or dict; Axes style parameters              palette="deep",				# string or sequence; Color palette              font="sans-serif", 		# Font family;               font_scale=1, 				# float, optional;               color_codes=True, 		# bool;              rc=None								# dict or None; matplotlib rc params             ):...
df = sns.load_dataset("penguins")plt.figure(dpi=300, figsize=(16, 16))plt.subplot(2, 2, 1)sns.barplot(data=df, x="island", y="body_mass_g")sns.set_theme()plt.subplot(2, 2, 2)sns.barplot(data=df, x="island", y="body_mass_g")sns.set_theme(style="whitegrid", palette="pastel")plt.subplot(2, 2, 3)sns.barplot(data=df, x="island", y="body_mass_g")custom_params = {"axes.spines.right": False, "axes.spines.top": False}sns.set_theme(style="ticks", rc=custom_params)plt.subplot(2, 2, 4)sns.barplot(data=df, x="island", y="body_mass_g")

image-20230829185843233

2.3.2 颜色主题
2.3.2.1 应用颜色主题

image-20230829190002618

df = sns.load_dataset("penguins")plt.figure(dpi=300, figsize=(16, 16))plt.subplot(2, 2, 1)sns.barplot(data=df, x="island", y="body_mass_g")sns.set_theme(palette="Accent")plt.subplot(2, 2, 2)sns.barplot(data=df, x="island", y="body_mass_g")sns.set_theme(palette="crest")plt.subplot(2, 2, 3)sns.barplot(data=df, x="island", y="body_mass_g")sns.set_theme(palette="blend:#7AB,#EDA")plt.subplot(2, 2, 4)sns.barplot(data=df, x="island", y="body_mass_g")

image-20230829190333883

2.3.2.2 使用 color_palette() 自定义色系
def color_palette(palette=None,                   # Name of palette or None to return current palette                  n_colors=None,                   # Number of colors in the palette                  desat=None,                   # Proportion to desaturate each color by                  as_cmap=False                  # If True, return a `matplotlib.colors.ListedColormap`                 ):...
sns.color_palette("ch:s=.25,rot=.25")				# cubehelix color palettesns.color_palette("light:#5A9")							# light sequential gradientsns.color_palette("dark:#5A9_r")						# [reversed] dark sequential gradientsns.color_palette("blend:#7AB,#EDA")				# blend gradient

image-20230829191148430

image-20230829191216420

image-20230829191236461

image-20230829191256697

2.3.3 绘图元素缩放比例

可以通过设置context参数调整绘图元素的相对大小,参数一共接受一下几种:

  • paper
  • notebook
  • talk
  • poster

绘图元素的缩放比例依次增大。

df = sns.load_dataset("penguins")plt.figure(dpi=300, figsize=(12, 12))plt.subplot(2, 2, 1)sns.barplot(data=df, x="island", y="body_mass_g")plt.title("Notebook")sns.set_theme(context="paper")plt.subplot(2, 2, 2)sns.barplot(data=df, x="island", y="body_mass_g")plt.title("Paper")sns.set_theme(context="talk")plt.subplot(2, 2, 3)sns.barplot(data=df, x="island", y="body_mass_g")plt.title("Talk")sns.set_theme(context="poster")plt.subplot(2, 2, 4)sns.barplot(data=df, x="island", y="body_mass_g")plt.title("Poster")plt.tight_layout()

image-20230829191948327