Matplotlib plt.scatter生成空白散点图问题求助

以下Python代码运行后返回空白散点图，但相同数据集与变量在R中使用geom_point()可正常绘图：

# import libraries
import pandas as pd
import os
import matplotlib.pyplot as plt
import numpy as np

os.chdir('file path')

# import data files
activity = pd.read_csv('file path\\dailyActivity_merged.csv')
intensity = pd.read_csv('file path\\hourlyIntensities_merged.csv')
steps = pd.read_csv('file path\\hourlySteps_merged.csv')
sleep = pd.read_csv('file path\\sleepDay_merged.csv')

# ActivityDate in activity df only includes dates (no time). Rename it Dates
activity = activity.rename(columns={'ActivityDate': 'Dates'})

# ActivityHour in intensity df and steps df includes date-time. Split date-time column into dates and times in intensity. Drop the date-time column

intensity['Dates'] = pd.to_datetime(intensity['ActivityHour']).dt.date
intensity['Times'] = pd.to_datetime(intensity['ActivityHour']).dt.time
intensity = intensity.drop(columns=['ActivityHour'])

# split date-time column into dates and times in steps. Drop the date-time column

steps['Dates'] = pd.to_datetime(steps['ActivityHour']).dt.date
steps['Times'] = pd.to_datetime(steps['ActivityHour']).dt.time
steps = steps.drop(columns=['ActivityHour'])

# split date-time column into dates and times in sleep. Drop the date-time column

sleep['Dates'] = pd.to_datetime(sleep['SleepDate']).dt.date
sleep['Times'] = pd.to_datetime(sleep['SleepDate']).dt.time
sleep = sleep.drop(columns=['SleepDate', 'TotalSleepRecords'])

# add a column & calculate time_awake_in_bed before falling asleep

sleep['time_awake_in_bed'] = sleep['TotalTimeInBed'] - sleep['TotalMinutesAsleep']

# merge activity and sleep
list = ['Id', 'Dates']
activity_sleep = sleep.merge(activity,
                on = list,
                how = 'outer')

# plot relation between calories used daily vs how long it takes users to fall asleep

plt.scatter(activity_sleep['time_awake_in_bed'], activity_sleep['Calories'], s=20, c='b', marker='o')
plt.axis([0, 200, 0, 5000])
plt.show()

注：max(Calories) = 4900，min(Calories) =0；修正笔误：max(time_awake_in_bed) = 150，min(time_awake_in_bed) = 0

问题根源 #

1. 日期类型不匹配导致合并失效：
   原代码仅重命名了activity的ActivityDate列，但未将其转换为datetime.date类型；而sleep的Dates是通过pd.to_datetime().dt.date生成的日期对象。两者类型不一致，outer join后大部分行无法匹配，产生大量含NaN的无效行。
2. 未过滤缺失值：
   合并后的activity_sleep中，time_awake_in_bed或Calories列存在大量NaN值，plt.scatter会自动忽略这些缺失值，导致无点可绘。

修复步骤及完整代码 #

1. 统一日期列类型 #

处理activity数据时，将ActivityDate转换为与sleep一致的datetime.date类型：

# 替换原activity重命名列的代码
activity['Dates'] = pd.to_datetime(activity['ActivityDate']).dt.date
activity = activity.drop(columns=['ActivityDate'])

2. 过滤合并后的无效数据 #

合并后筛选出time_awake_in_bed和Calories均不为空的行：

# 合并数据后添加过滤步骤
activity_sleep = activity_sleep.dropna(subset=['time_awake_in_bed', 'Calories'])

完整修正代码 #

# import libraries
import pandas as pd
import os
import matplotlib.pyplot as plt
import numpy as np

os.chdir('file path')

# import data files
activity = pd.read_csv('file path\\dailyActivity_merged.csv')
intensity = pd.read_csv('file path\\hourlyIntensities_merged.csv')
steps = pd.read_csv('file path\\hourlySteps_merged.csv')
sleep = pd.read_csv('file path\\sleepDay_merged.csv')

# 统一activity的日期格式为datetime.date
activity['Dates'] = pd.to_datetime(activity['ActivityDate']).dt.date
activity = activity.drop(columns=['ActivityDate'])

# intensity数据处理（原代码无需修改）
intensity['Dates'] = pd.to_datetime(intensity['ActivityHour']).dt.date
intensity['Times'] = pd.to_datetime(intensity['ActivityHour']).dt.time
intensity = intensity.drop(columns=['ActivityHour'])

# steps数据处理（原代码无需修改）
steps['Dates'] = pd.to_datetime(steps['ActivityHour']).dt.date
steps['Times'] = pd.to_datetime(steps['ActivityHour']).dt.time
steps = steps.drop(columns=['ActivityHour'])

# sleep数据处理
sleep['Dates'] = pd.to_datetime(sleep['SleepDate']).dt.date
sleep['Times'] = pd.to_datetime(sleep['SleepDate']).dt.time
sleep = sleep.drop(columns=['SleepDate', 'TotalSleepRecords'])

# 计算醒着的时间
sleep['time_awake_in_bed'] = sleep['TotalTimeInBed'] - sleep['TotalMinutesAsleep']

# 合并数据并过滤无效行
activity_sleep = sleep.merge(activity, on=['Id', 'Dates'], how='outer')
activity_sleep = activity_sleep.dropna(subset=['time_awake_in_bed', 'Calories'])

# 绘制散点图
plt.scatter(activity_sleep['time_awake_in_bed'], activity_sleep['Calories'], s=20, c='b', marker='o')
plt.axis([0, 200, 0, 5000])
plt.xlabel('Time Awake in Bed')
plt.ylabel('Calories')
plt.title('Calories vs Time Awake in Bed')
plt.show()

验证建议 #

绘图前可打印数据基本信息，确认有效数据量：

print(activity_sleep[['time_awake_in_bed', 'Calories']].info())
print(activity_sleep[['time_awake_in_bed', 'Calories']].describe())

内容的提问来源于stack exchange，提问作者Bashir