Python matplotlib实战之漏斗图绘制

fp = "d:/share/data/A0M0203.csv"
df = pd.read_csv(fp)
df

data = df[df["sj"] == 2021]
#A0M020312: 普通小学毕业生数(万人)
#A0M02030T: 初中阶段教育毕业生数(万人)	
#A0M02030J: 普通高中毕业生数(万人)
#A0M020306: 普通本科毕业生数(万人)	
#A0M020304: 硕士毕业生数(万人)	
#A0M020303: 博士毕业生数(万人)
data = data[
    data["zb"].isin(
        [
            "A0M020312",
            "A0M02030T",
            "A0M02030J",
            "A0M020306",
            "A0M020304",
            "A0M020303",
        ]
    )
]
data = data.sort_values("value", ascending=False)
data

with plt.style.context("dark_background"):
    fig = plt.figure()
    ax = fig.add_axes([0.1, 0.1, 1, 1])
    colors = plt.cm.Set2.colors
    cnt = len(data)
    y = [[1 + i * 3, 3.8 + i * 3] for i in range(cnt)]
    y_ticks = [2 + i * 3 for i in range(cnt)]
    start_x1 = 5
    start_x2 = -5
    for i in range(cnt):
        ax.fill_betweenx(
            y=y[i],
            x1=[start_x1, data.iloc[i, 4]],
            x2=[start_x2, -1 * data.iloc[i, 4]],
            color=colors[i],
        )
        start_x1 = data.iloc[i, 4]
        start_x2 = -1 * data.iloc[i, 4]
    ax.set_xticks([], [])
    ax.set_yticks(y_ticks, data["zbCN"])
    for y, value in zip(y_ticks, data["value"]):
        ax.text(
            10,
            y,
            value,
            fontsize=16,
            fontweight="bold",
            color="white",
            ha="center",
        )
    ax.grid(False)
    ax.set_title("2021年各学历毕业人数")