Details for fitbit_sleep_analysis.ipynb
Published by Wenqiu999
Description
Analysis of sleep data collected by Fitbit.
0
Tags & Data Sources
fitbit connection sleep fitbit connectionComments
Please log in to comment.Notebook Last updated 4 years, 5 months ago
This notebook is created to analyze sleep data collected by fitbit. It used public data as a demonstrate to show how the plots look like.¶
You can use the code below to analyze your own sleep data. And This notebook is free to reuse and adapt, distributed under an MIT license: https://opensource.org/licenses/MIT¶
You can choose the year you want to analyze.¶
There might be some days when your device failed to collect your data. Let's remove the dates that data were not collected.¶
First, by calculatting the average sleep hours and the standard deviation, you can know whether you are getting the required hours of sleep?¶
Out[8]:
The value of mean tells you your average sleeping hours and the standard deviation can tell your how your sleeping data spread out. The smaller the standard deviation is, the more centralized your data is. Then, let's use a bar chart to explore the distribution of your sleeping hours.¶
From the bar chart above, you can learn about how many hours you sleep most of the time. Let's also plot your sleeping hours through the whole year with the average line.¶
When you're in bed, you're not asleep all the time. Let's take a look at the sleep pattern while you're in bed.¶
Now, let's take a look at whether there is any pattern of your sleep based on different days of a week.¶
Out[14]:
Let's check out how your sleep varies for weekdays vs weekends.¶
Out[17]:
Camparing the boxes above, you can learn about how your sleep varies during a week. And the change of seasons may affect your sleep,too. Let's how your sleep changes over month in one year.¶
This notebook is created to analyze sleep data collected by fitbit. It used public data as a demonstrate to show how the plots look like.¶
You can use the code below to analyze your own sleep data. And This notebook is free to reuse and adapt, distributed under an MIT license: https://opensource.org/licenses/MIT¶
You can choose the year you want to analyze.¶
There might be some days when your device failed to collect your data. Let's remove the dates that data were not collected.¶
First, by calculatting the average sleep hours and the standard deviation, you can know whether you are getting the required hours of sleep?¶
Out[8]:
The value of mean tells you your average sleeping hours and the standard deviation can tell your how your sleeping data spread out. The smaller the standard deviation is, the more centralized your data is. Then, let's use a bar chart to explore the distribution of your sleeping hours.¶
From the bar chart above, you can learn about how many hours you sleep most of the time. Let's also plot your sleeping hours through the whole year with the average line.¶
When you're in bed, you're not asleep all the time. Let's take a look at the sleep pattern while you're in bed.¶
Now, let's take a look at whether there is any pattern of your sleep based on different days of a week.¶
Out[14]:
Let's check out how your sleep varies for weekdays vs weekends.¶
Out[17]:
Camparing the boxes above, you can learn about how your sleep varies during a week. And the change of seasons may affect your sleep,too. Let's how your sleep changes over month in one year.¶
Notebook Last updated 4 years, 5 months ago
This notebook is created to analyze sleep data collected by fitbit. It used public data as a demonstrate to show how the plots look like.¶
You can use the code below to analyze your own sleep data. And This notebook is free to reuse and adapt, distributed under an MIT license: https://opensource.org/licenses/MIT¶
In [1]:
import pandas as pd
import json
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
import urllib
import requests
import os
import tempfile
from datetime import datetime
from ohapi import api
user_details = api.exchange_oauth2_member(os.environ.get('OH_ACCESS_TOKEN'))
for i in user_details['data']:
if i['basename'] == 'fitbit-data.json' and i['source'] == 'direct-sharing-102':
fitbit = requests.get(i['download_url']).json()
'''
data = requests.get('https://www.openhumans.org/api/public/project/102/datafiles/').json()
for i in data['results']:
if i['id'] == 9135000:
fitbit = requests.get(i['download_url']).json()
'''
In [2]:
DateTime = []
You can choose the year you want to analyze.¶
In [3]:
year = '2018'
In [4]:
def GetData(obj,year,key):
name = key.split('-')[-1]
name = []
for day in fitbit[obj][year][key]:
name.append(day['value'])
return name
for day in fitbit['minutes-to-sleep'][year]['sleep-minutesToFallAsleep']:
DateTime.append(datetime.strptime(day['dateTime'],'%Y-%m-%d'))
sleepData = pd.DataFrame(
data = {
'minutesToFallAsleep': GetData('minutes-to-sleep',year,'sleep-minutesToFallAsleep'),
'minutesAfterWakeup': GetData('sleep-minutes-after-wakeup',year,'sleep-minutesAfterWakeup'),
'startTime': GetData('sleep-start-time',year,'sleep-startTime'),
'minutesAwake': GetData('awake-minutes',year,'sleep-minutesAwake'),
'timeInBed': GetData('time-in-bed',year,'sleep-timeInBed'),
'awakeningsCount': GetData('sleep-awakenings',year,'sleep-awakeningsCount'),
'sleepEfficiency': GetData('sleep-efficiency',year,'sleep-efficiency'),
'minutesAsleep':GetData('sleep-minutes',year,'sleep-minutesAsleep')
}
)
In [5]:
DateTime = pd.DataFrame(DateTime)
In [6]:
sleepData = pd.concat([DateTime, sleepData], axis=1)
sleepData.rename(columns={sleepData.columns[0]: 'DateTime'}, inplace=True)
sleepData = pd.DataFrame(sleepData.set_index('DateTime', drop= False))
There might be some days when your device failed to collect your data. Let's remove the dates that data were not collected.¶
In [7]:
col = ['minutesToFallAsleep','minutesAfterWakeup','timeInBed','awakeningsCount','minutesAsleep','sleepEfficiency']
sleepData[col] = sleepData[col].mask(sleepData[col].applymap(str).eq('0'))
First, by calculatting the average sleep hours and the standard deviation, you can know whether you are getting the required hours of sleep?¶
In [8]:
import matplotlib.dates as mdates
sleepData['timeInBed'] = sleepData['timeInBed'].astype(float)
sleepData['hourInBed'] = sleepData['timeInBed']/60
sleepDesc = pd.DataFrame(sleepData['hourInBed'].describe())
sleepDesc
Out[8]:
The value of mean tells you your average sleeping hours and the standard deviation can tell your how your sleeping data spread out. The smaller the standard deviation is, the more centralized your data is. Then, let's use a bar chart to explore the distribution of your sleeping hours.¶
In [9]:
avgSleepHours = round(sleepDesc['hourInBed']['mean'],2)
summary = 'Averaging a sleep of {} hours with a standardized deviation of {} hours'.format(avgSleepHours, round(sleepDesc['hourInBed']['std'],2))
fig, ax = plt.subplots(figsize = (20,6))
plt.hist(sleepData['hourInBed'], bins = 8, range = (3, 10), color="navy")
plt.xlim(3, 10)
plt.xticks(range(3, 10))
plt.axvline(avgSleepHours, color="orangered", linestyle='--')
plt.xlabel('Hours in Bed')
plt.ylabel('Count');
plt.title('Figure 1. Bar chart of Hours in Bed', fontsize=15)
bbox_props1 = dict(boxstyle='round4, pad=0.6', fc='cyan', ec='b', lw=.5)
ax.annotate(summary,
fontsize=12,
fontweight='demi',
xy=(avgSleepHours, 120),
xycoords='data',
xytext=(10, -30),
textcoords='offset points',
arrowprops=dict(arrowstyle="simple",), bbox=bbox_props1)
plt.show()
From the bar chart above, you can learn about how many hours you sleep most of the time. Let's also plot your sleeping hours through the whole year with the average line.¶
In [10]:
fig, ax = plt.subplots(figsize = (20,6))
plt.plot(sleepData['DateTime'],sleepData['hourInBed'], linestyle='-',
markersize=10, color='c', label='% Light', linewidth=3.0, alpha=0.9)
plt.ylabel('hourInBed', fontsize=14)
plt.axhline(avgSleepHours, color="orangered", linestyle='--')
plt.title('Figure 2. Line plot of Hours in Bed over {}'.format(year), fontsize = 15)
ax.annotate
ax.xaxis.set_major_locator(mdates.WeekdayLocator(byweekday=6))
ax.xaxis.set_major_formatter(mdates.DateFormatter('%D'))
ax.grid(True)
plt.xticks(rotation=90)
plt.plot()
plt.show()
When you're in bed, you're not asleep all the time. Let's take a look at the sleep pattern while you're in bed.¶
In [11]:
SleepPattern = sleepData[['minutesAwake','minutesAsleep']].astype(float)
avgSleep = SleepPattern.mean()
In [12]:
fig = plt.figure(figsize = (6,6))
labels=['Awake', 'Asleep']
plt.pie(avgSleep, colors = ['lightskyblue', 'yellowgreen'], autopct='%1.1f%%', labels=labels, textprops=dict(color="w"))
my_circle=plt.Circle( (0,0), 0.7, color='white')
p=plt.gcf()
p.gca().add_artist(my_circle)
plt.title('Figure 3. Average of types of sleep over {}'.format(year), fontsize=15)
plt.legend()
plt.show()
Now, let's take a look at whether there is any pattern of your sleep based on different days of a week.¶
In [13]:
sleepData['Day of week'] = sleepData.index.weekday_name
weekday = ['Monday','Tuesday','Wednesday','Thursday','Friday']
weekend = ['Saturday','Sunday']
sleepData.loc[sleepData['Day of week'].isin(weekday), 'IsWeekday'] = 'Weekday'
sleepData.loc[sleepData['Day of week'].isin(weekend), 'IsWeekday'] = 'Weekend'
In [14]:
sleepData[['minutesAsleep', 'minutesAwake', 'sleepEfficiency']] = sleepData[['minutesAsleep', 'minutesAwake', 'sleepEfficiency']].astype(float)
dayGroupedData = sleepData.groupby(['Day of week']).mean()
dayGroupedData
Out[14]:
In [15]:
dayofweek = ['Mon','Tue','Wed','Thu','Fri','Sat','Sun']
fig = plt.figure(figsize = (10,5))
plt.bar((dayGroupedData.index), dayGroupedData['minutesAwake'], width = 0.4, color='slateblue', label='Minutes Awake')
plt.bar((dayGroupedData.index), dayGroupedData['minutesAsleep'], width = 0.4, color='dodgerblue', label='Minutes Asleep', bottom=dayGroupedData['minutesAwake'])
plt.bar((dayGroupedData.index), dayGroupedData['timeInBed'],width = 0.4, color='lightskyblue', label='Time in Bed', bottom=dayGroupedData['minutesAsleep'])
plt.xticks(np.arange(len(dayofweek)),dayofweek)
plt.title('Figure 4. Wake/Asleep Minutes While in Bed over One Week ', fontsize = 15)
plt.legend(loc='upper right')
plt.show()
In [16]:
fig, ax = plt.subplots(figsize=(10, 5))
ax.plot('Day of week', 'timeInBed', data=sleepData, marker='o', markersize=8, linestyle='None', color ='salmon')
dayGroupedData = dayGroupedData.reset_index(drop=False)
ax.plot(dayGroupedData['timeInBed'],color='r', linewidth=2, label='Average Time in Bed' )
ax.set_ylabel('Time in Bed')
ax.set_xlabel('Day of Week')
plt.title('Figure 5. Run Chart of Minutes in Bed with Average value by Day of the Week', fontsize = 15)
plt.xticks(np.arange(len(dayofweek)),dayofweek)
plt.legend()
plt.show()
Let's check out how your sleep varies for weekdays vs weekends.¶
In [17]:
sleepData.groupby('IsWeekday').mean()
Out[17]:
In [18]:
fig, ((ax1, ax2), (ax3, ax4)) = plt.subplots(ncols=2,nrows=2,figsize=(10, 10))
xticks = ['Weekday','Weekend']
sleepData.groupby('IsWeekday')['timeInBed'].mean().plot(kind='bar', ax = ax1)
ax1.set_title('Comparison of Time in Bed')
ax1.set_xticklabels(xticks, rotation = 0, ha="right")
ax1.xaxis.label.set_visible(False)
sleepData.groupby('IsWeekday')['minutesAsleep'].mean().plot(kind='bar', ax = ax2)
ax2.set_title('Comparison of Minutes Asleep')
ax2.set_xticklabels(xticks, rotation = 0, ha="right")
ax2.xaxis.label.set_visible(False)
sleepData.groupby('IsWeekday')['minutesAwake'].mean().plot(kind='bar', ax = ax3)
ax3.set_title('Comparison of Minutes Awake')
ax3.set_xticklabels(xticks,rotation = 0, ha="right")
ax3.xaxis.label.set_visible(False)
sleepData.groupby('IsWeekday')['sleepEfficiency'].mean().plot(kind='bar', ax = ax4)
ax4.set_title('Comparison of Sleep Efficiency')
ax4.set_xticklabels(xticks,rotation = 0, ha="right")
ax4.xaxis.label.set_visible(False)
plt.suptitle('Figure 6. Sleep Comparison Between Weekday and Weekend')
plt.show()
Camparing the boxes above, you can learn about how your sleep varies during a week. And the change of seasons may affect your sleep,too. Let's how your sleep changes over month in one year.¶
In [19]:
sleepData['Month'] = sleepData.index.month
In [20]:
fig, axes = plt.subplots(3, 1, figsize=(12,10), sharex=False)
Month = ['Jan','Feb','Mar','Apr','May','Jun','Jul','Aug','Sep','Oct','Nov','Dec']
for name, ax in zip(['timeInBed','minutesAsleep', 'minutesAwake','sleepEfficiency'], axes):
sns.boxplot(data=sleepData, x='Month', y=name, ax=ax)
ax.set_ylabel('Minutes')
ax.set_title(name)
ax.set_xticklabels(Month)
if ax != axes[-1]:
ax.set_xlabel('')
plt.suptitle('Figure 7. Sleep Change over Year {}'.format(year))
plt.show()
In [ ]:
In [ ]:
This notebook is created to analyze sleep data collected by fitbit. It used public data as a demonstrate to show how the plots look like.¶
You can use the code below to analyze your own sleep data. And This notebook is free to reuse and adapt, distributed under an MIT license: https://opensource.org/licenses/MIT¶
In [1]:
import pandas as pd
import json
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
import urllib
import requests
import os
import tempfile
from datetime import datetime
from ohapi import api
user_details = api.exchange_oauth2_member(os.environ.get('OH_ACCESS_TOKEN'))
for i in user_details['data']:
if i['basename'] == 'fitbit-data.json' and i['source'] == 'direct-sharing-102':
fitbit = requests.get(i['download_url']).json()
'''
data = requests.get('https://www.openhumans.org/api/public/project/102/datafiles/').json()
for i in data['results']:
if i['id'] == 9135000:
fitbit = requests.get(i['download_url']).json()
'''
In [2]:
DateTime = []
You can choose the year you want to analyze.¶
In [3]:
year = '2018'
In [4]:
def GetData(obj,year,key):
name = key.split('-')[-1]
name = []
for day in fitbit[obj][year][key]:
name.append(day['value'])
return name
for day in fitbit['minutes-to-sleep'][year]['sleep-minutesToFallAsleep']:
DateTime.append(datetime.strptime(day['dateTime'],'%Y-%m-%d'))
sleepData = pd.DataFrame(
data = {
'minutesToFallAsleep': GetData('minutes-to-sleep',year,'sleep-minutesToFallAsleep'),
'minutesAfterWakeup': GetData('sleep-minutes-after-wakeup',year,'sleep-minutesAfterWakeup'),
'startTime': GetData('sleep-start-time',year,'sleep-startTime'),
'minutesAwake': GetData('awake-minutes',year,'sleep-minutesAwake'),
'timeInBed': GetData('time-in-bed',year,'sleep-timeInBed'),
'awakeningsCount': GetData('sleep-awakenings',year,'sleep-awakeningsCount'),
'sleepEfficiency': GetData('sleep-efficiency',year,'sleep-efficiency'),
'minutesAsleep':GetData('sleep-minutes',year,'sleep-minutesAsleep')
}
)
In [5]:
DateTime = pd.DataFrame(DateTime)
In [6]:
sleepData = pd.concat([DateTime, sleepData], axis=1)
sleepData.rename(columns={sleepData.columns[0]: 'DateTime'}, inplace=True)
sleepData = pd.DataFrame(sleepData.set_index('DateTime', drop= False))
There might be some days when your device failed to collect your data. Let's remove the dates that data were not collected.¶
In [7]:
col = ['minutesToFallAsleep','minutesAfterWakeup','timeInBed','awakeningsCount','minutesAsleep','sleepEfficiency']
sleepData[col] = sleepData[col].mask(sleepData[col].applymap(str).eq('0'))
First, by calculatting the average sleep hours and the standard deviation, you can know whether you are getting the required hours of sleep?¶
In [8]:
import matplotlib.dates as mdates
sleepData['timeInBed'] = sleepData['timeInBed'].astype(float)
sleepData['hourInBed'] = sleepData['timeInBed']/60
sleepDesc = pd.DataFrame(sleepData['hourInBed'].describe())
sleepDesc
Out[8]:
The value of mean tells you your average sleeping hours and the standard deviation can tell your how your sleeping data spread out. The smaller the standard deviation is, the more centralized your data is. Then, let's use a bar chart to explore the distribution of your sleeping hours.¶
In [9]:
avgSleepHours = round(sleepDesc['hourInBed']['mean'],2)
summary = 'Averaging a sleep of {} hours with a standardized deviation of {} hours'.format(avgSleepHours, round(sleepDesc['hourInBed']['std'],2))
fig, ax = plt.subplots(figsize = (20,6))
plt.hist(sleepData['hourInBed'], bins = 8, range = (3, 10), color="navy")
plt.xlim(3, 10)
plt.xticks(range(3, 10))
plt.axvline(avgSleepHours, color="orangered", linestyle='--')
plt.xlabel('Hours in Bed')
plt.ylabel('Count');
plt.title('Figure 1. Bar chart of Hours in Bed', fontsize=15)
bbox_props1 = dict(boxstyle='round4, pad=0.6', fc='cyan', ec='b', lw=.5)
ax.annotate(summary,
fontsize=12,
fontweight='demi',
xy=(avgSleepHours, 120),
xycoords='data',
xytext=(10, -30),
textcoords='offset points',
arrowprops=dict(arrowstyle="simple",), bbox=bbox_props1)
plt.show()
From the bar chart above, you can learn about how many hours you sleep most of the time. Let's also plot your sleeping hours through the whole year with the average line.¶
In [10]:
fig, ax = plt.subplots(figsize = (20,6))
plt.plot(sleepData['DateTime'],sleepData['hourInBed'], linestyle='-',
markersize=10, color='c', label='% Light', linewidth=3.0, alpha=0.9)
plt.ylabel('hourInBed', fontsize=14)
plt.axhline(avgSleepHours, color="orangered", linestyle='--')
plt.title('Figure 2. Line plot of Hours in Bed over {}'.format(year), fontsize = 15)
ax.annotate
ax.xaxis.set_major_locator(mdates.WeekdayLocator(byweekday=6))
ax.xaxis.set_major_formatter(mdates.DateFormatter('%D'))
ax.grid(True)
plt.xticks(rotation=90)
plt.plot()
plt.show()
When you're in bed, you're not asleep all the time. Let's take a look at the sleep pattern while you're in bed.¶
In [11]:
SleepPattern = sleepData[['minutesAwake','minutesAsleep']].astype(float)
avgSleep = SleepPattern.mean()
In [12]:
fig = plt.figure(figsize = (6,6))
labels=['Awake', 'Asleep']
plt.pie(avgSleep, colors = ['lightskyblue', 'yellowgreen'], autopct='%1.1f%%', labels=labels, textprops=dict(color="w"))
my_circle=plt.Circle( (0,0), 0.7, color='white')
p=plt.gcf()
p.gca().add_artist(my_circle)
plt.title('Figure 3. Average of types of sleep over {}'.format(year), fontsize=15)
plt.legend()
plt.show()
Now, let's take a look at whether there is any pattern of your sleep based on different days of a week.¶
In [13]:
sleepData['Day of week'] = sleepData.index.weekday_name
weekday = ['Monday','Tuesday','Wednesday','Thursday','Friday']
weekend = ['Saturday','Sunday']
sleepData.loc[sleepData['Day of week'].isin(weekday), 'IsWeekday'] = 'Weekday'
sleepData.loc[sleepData['Day of week'].isin(weekend), 'IsWeekday'] = 'Weekend'
In [14]:
sleepData[['minutesAsleep', 'minutesAwake', 'sleepEfficiency']] = sleepData[['minutesAsleep', 'minutesAwake', 'sleepEfficiency']].astype(float)
dayGroupedData = sleepData.groupby(['Day of week']).mean()
dayGroupedData
Out[14]:
In [15]:
dayofweek = ['Mon','Tue','Wed','Thu','Fri','Sat','Sun']
fig = plt.figure(figsize = (10,5))
plt.bar((dayGroupedData.index), dayGroupedData['minutesAwake'], width = 0.4, color='slateblue', label='Minutes Awake')
plt.bar((dayGroupedData.index), dayGroupedData['minutesAsleep'], width = 0.4, color='dodgerblue', label='Minutes Asleep', bottom=dayGroupedData['minutesAwake'])
plt.bar((dayGroupedData.index), dayGroupedData['timeInBed'],width = 0.4, color='lightskyblue', label='Time in Bed', bottom=dayGroupedData['minutesAsleep'])
plt.xticks(np.arange(len(dayofweek)),dayofweek)
plt.title('Figure 4. Wake/Asleep Minutes While in Bed over One Week ', fontsize = 15)
plt.legend(loc='upper right')
plt.show()
In [16]:
fig, ax = plt.subplots(figsize=(10, 5))
ax.plot('Day of week', 'timeInBed', data=sleepData, marker='o', markersize=8, linestyle='None', color ='salmon')
dayGroupedData = dayGroupedData.reset_index(drop=False)
ax.plot(dayGroupedData['timeInBed'],color='r', linewidth=2, label='Average Time in Bed' )
ax.set_ylabel('Time in Bed')
ax.set_xlabel('Day of Week')
plt.title('Figure 5. Run Chart of Minutes in Bed with Average value by Day of the Week', fontsize = 15)
plt.xticks(np.arange(len(dayofweek)),dayofweek)
plt.legend()
plt.show()
Let's check out how your sleep varies for weekdays vs weekends.¶
In [17]:
sleepData.groupby('IsWeekday').mean()
Out[17]:
In [18]:
fig, ((ax1, ax2), (ax3, ax4)) = plt.subplots(ncols=2,nrows=2,figsize=(10, 10))
xticks = ['Weekday','Weekend']
sleepData.groupby('IsWeekday')['timeInBed'].mean().plot(kind='bar', ax = ax1)
ax1.set_title('Comparison of Time in Bed')
ax1.set_xticklabels(xticks, rotation = 0, ha="right")
ax1.xaxis.label.set_visible(False)
sleepData.groupby('IsWeekday')['minutesAsleep'].mean().plot(kind='bar', ax = ax2)
ax2.set_title('Comparison of Minutes Asleep')
ax2.set_xticklabels(xticks, rotation = 0, ha="right")
ax2.xaxis.label.set_visible(False)
sleepData.groupby('IsWeekday')['minutesAwake'].mean().plot(kind='bar', ax = ax3)
ax3.set_title('Comparison of Minutes Awake')
ax3.set_xticklabels(xticks,rotation = 0, ha="right")
ax3.xaxis.label.set_visible(False)
sleepData.groupby('IsWeekday')['sleepEfficiency'].mean().plot(kind='bar', ax = ax4)
ax4.set_title('Comparison of Sleep Efficiency')
ax4.set_xticklabels(xticks,rotation = 0, ha="right")
ax4.xaxis.label.set_visible(False)
plt.suptitle('Figure 6. Sleep Comparison Between Weekday and Weekend')
plt.show()
Camparing the boxes above, you can learn about how your sleep varies during a week. And the change of seasons may affect your sleep,too. Let's how your sleep changes over month in one year.¶
In [19]:
sleepData['Month'] = sleepData.index.month
In [20]:
fig, axes = plt.subplots(3, 1, figsize=(12,10), sharex=False)
Month = ['Jan','Feb','Mar','Apr','May','Jun','Jul','Aug','Sep','Oct','Nov','Dec']
for name, ax in zip(['timeInBed','minutesAsleep', 'minutesAwake','sleepEfficiency'], axes):
sns.boxplot(data=sleepData, x='Month', y=name, ax=ax)
ax.set_ylabel('Minutes')
ax.set_title(name)
ax.set_xticklabels(Month)
if ax != axes[-1]:
ax.set_xlabel('')
plt.suptitle('Figure 7. Sleep Change over Year {}'.format(year))
plt.show()
In [ ]:
In [ ]: