主页 / 问题 / 1217640
Accepted
Владислав Харламов
Asked:2021-12-12 15:00:53 +0000 UTC2021-12-12 15:00:53 +0000 UTC

在 seaborn 线图中跳过缺失数据

  • 772

在此处输入图像描述

上面有一个图表。我想尝试从这些数据中重现它。

我这样做：

def month2season(month):
    if 1 <= month <= 2 or month == 12:
        return 'winter'
    elif 3 <= month <= 5:
        return 'spring'
    elif 6 <= month <= 8:
        return 'summer'
    elif 9 <= month <= 11:
        return 'autumn'

df = pd.read_csv('https://raw.githubusercontent.com/ternikov/hse/gh-pages/hw5/temp_data.csv')
df_grouped = df.groupby(['year', 'month'], as_index=False).mean()
df_grouped['season'] = df_grouped['month'].apply(month2season)

fig, ax = plt.subplots(2, 2, figsize=(15, 10))
for season_data, axes in zip(df_grouped['season'].unique(), ax.reshape(-1)):
    season_data = df_grouped[df_grouped['season'] == season_data]
    sns.lineplot('year', 'temp', data=season_data, ax=axes)
plt.show()

但问题是它lineplot以这种方式连接了具有缺失值的区域：

在此处输入图像描述

如何纠正？

python

1 个回答

  1. Best Answer

    经过一些实验，事实证明 Seaborn 丢弃了无穷大值np.inf（not NaN

    def month2season(m):
    seasons = ["Winter", "Spring", "Summer", "Autumn"]
    return np.take(seasons, (m % 12) // 3)

df = pd.read_csv("https://raw.githubusercontent.com/ternikov/hse/gh-pages/hw5/temp_data.csv")
int_cols = ["year", "month"]
df[int_cols] = df[int_cols].astype(int)

res = pd.DataFrame(list(product(range(1952, 2017), range(1, 13))), columns=int_cols)
res = res.merge(df, how="left")
res["season"] = res["month"].map(month2season)

fig, axes = plt.subplots(2, 2, figsize=(15, 10))

for season, ax in zip(res['season'].unique(), axes.ravel()):
    sns.lineplot('year', 'temp', data=res[res['season'] == season].fillna(np.inf), ax=ax)

    在此处输入图像描述

    • 6

相关问题