Spotify Data Analysis and Visualization

Note: The data presented below are from 6 January 2021 to 21 September 2021.

Making the Imports

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
from plotly import __version__
from plotly.offline import download_plotlyjs, init_notebook_mode, plot, iplot
import cufflinks as cf
init_notebook_mode(connected=True)
cf.go_offline()
import plotly.graph_objects as go
import plotly.express as px

Loading The DataFrames

songs_by_time = pd.read_csv('MyData\songs_by_time.csv')
songs_raw = pd.read_csv('MyData\songs_raw.csv')
songs_by_artists = pd.read_csv('MyData\songs_by_artists.csv')

1. What Fraction of Time is Spent on Listening to Spotify?

#Total Time
total_time_minutes= songs_raw['Minutes Played'].sum()
total_time_days = total_time_minutes/(60*24)

#Total Days of listening to Spotify
first_date = pd.to_datetime(songs_raw['Date'].iloc[0])
last_date =pd.to_datetime(songs_raw['Date'].iloc[-1])
number_of_days = (last_date-first_date).days

#Perecentage Calculation
percent = (total_time_days/number_of_days)*100

#Result
print(f"I've spent {np.round(percent, 2)}% of my time on Spotify!")

I've spent 11.01% of my time on Spotify!

Result

It turns out I have used 11.01% time of my day listening to Spotify. That is a 2.5 hours daily average!

2. Number of Songs and Total Songs Played

total_played = songs_raw.shape[0]
start_date = songs_raw['Date'].iloc[0]
final_date = songs_raw['Date'].iloc[-1]
print(f"I've played {total_played} songs from {start_date} to {final_date}!")

I've played 13550 songs from 06 Jan 2021 to 21 Sep 2021!

indi_songs = len(songs_raw.groupby('Song').sum())
print(f"I've played {indi_songs} different songs from {start_date} to {final_date}!")

I've played 599 different songs from 06 Jan 2021 to 21 Sep 2021!

Result

A. Total Songs Played = 13550
B Individual Songs Played = 599
This means that on average a song has been played about 22 times!

3. Day Wise Summary

#Day Wise Usage
day_wise = songs_raw.groupby('Date').sum().sort_values(by='Minutes Played', ascending=False)
day_wise['Hours Played'] = day_wise['Minutes Played']/60

#Rounding the numbers to two decimal places
day_wise['Hours Played'] = day_wise['Hours Played'].apply(lambda x: np.round(x,2))
day_wise['Minutes Played'] = day_wise['Minutes Played'].apply(lambda x: np.round(x,2))

#Top 10 Days with maximum number of hours
day_wise.head(10)

	Minutes Played	Hours Played
Date
20 Feb 2021	496.43	8.27
23 Jan 2021	462.33	7.71
06 Feb 2021	457.45	7.62
23 Mar 2021	448.84	7.48
25 Jan 2021	446.96	7.45
25 Feb 2021	444.57	7.41
18 Jan 2021	439.94	7.33
08 Jan 2021	412.15	6.87
21 Feb 2021	394.75	6.58
17 Mar 2021	376.28	6.27

#Bottom 10 Days with maximum number of hours
day_wise.tail(10)

	Minutes Played	Hours Played
Date
27 May 2021	31.78	0.53
12 Jul 2021	27.92	0.47
26 Aug 2021	26.79	0.45
20 Jun 2021	26.25	0.44
28 Apr 2021	24.98	0.42
23 Aug 2021	24.77	0.41
28 May 2021	24.47	0.41
02 May 2021	23.22	0.39
30 Apr 2021	18.33	0.31
23 Apr 2021	17.57	0.29

#Making a plot showing Day Wise Streaming Time in minutes and Day
day_wise['Minutes Played'].iplot(color='red',
title='Day Wise Streaming Time', yTitle = 'Number of Minutes', xTitle = 'Days', width = 3)

#Making a plot showing Day Wise Streaming Time in Hours and Day
day_wise['Hours Played'].iplot(color='green',
title='Day Wise Streaming Time', yTitle = 'Number of Hours', xTitle = 'Days', width = 3)

day_wise.to_csv('MyData\day_wise.csv')

day_wise=pd.read_csv('MyData\day_wise.csv')

index = pd.to_datetime(day_wise['Date']).sort_values().index

day_wise_sorted = day_wise.reindex(index)

#Making a barplot showing Day Wise Streaming Time and that date
day_wise_sorted.iplot(kind = 'bar', color='green', x= 'Date', y='Hours Played',
title='Day Wise Streaming Time', yTitle = 'Number of Hours', xTitle = 'Days')

#Making a lineplot showing Day Wise Streaming Time in Hours and that date
day_wise_sorted.iplot(kind = 'line', color='green', x= 'Date', y='Hours Played',
title='Day Wise Streaming Time', yTitle = 'Number of Hours', xTitle = 'Days', width = 3)

#Making a barplot showing Day Wise Streaming Time and that date
day_wise_sorted.iplot(kind = 'bar', color='blue', x= 'Date', y='Minutes Played',
title='Day Wise Streaming Time', yTitle = 'Number of Minutes', xTitle = 'Days')

#Making a lineplot showing Day Wise Streaming Time in Hours and that date
day_wise_sorted.iplot(kind = 'line', color='blue', x= 'Date', y='Minutes Played',
title='Day Wise Streaming Time', yTitle = 'Number of Minutes', xTitle = 'Days', width = 2)

Result

1. At my prime days I've used Spotify upto 8 hours in a day.
   
2. At my worst days the time has reduced to about half an hour.

4. Favourite Artists

# Making Hours Played column
songs_by_artists['Hours Played'] = songs_by_artists['Minutes Played']/60

#Making a percentage column
total_time = songs_by_artists['Minutes Played'].sum()
songs_by_artists['Percent of Overall Time (in %)']= songs_by_artists['Minutes Played'].apply(lambda x:np.round(x*100/total_time, 2))

#Rounding the numbers to two decimal places
songs_by_artists['Minutes Played'] = songs_by_artists['Minutes Played'].apply(lambda x: np.round(x,2))
songs_by_artists['Hours Played'] = songs_by_artists['Hours Played'].apply(lambda x: np.round(x,2))

# Top 10 Artists
songs_by_artists.head(10)

	Artist	Minutes Played	Hours Played	Percent of Overall Time (in %)
0	Taylor Swift	19419.25	323.65	47.61
1	Céline Dion	5278.13	87.97	12.94
2	Ed Sheeran	3249.73	54.16	7.97
3	Shawn Mendes	2141.92	35.70	5.25
4	Alan Walker	2025.89	33.76	4.97
5	Lady Gaga	1183.59	19.73	2.90
6	Maren Morris	1129.92	18.83	2.77
7	Camila Cabello	1067.44	17.79	2.62
8	Dua Lipa	819.79	13.66	2.01
9	Christina Perri	666.50	11.11	1.63

#Making a plot showing Artist Wise Streaming Time in Minutes
songs_by_artists.iplot(color='green', x= 'Artist', y = 'Minutes Played',
title='Minutes Played and Artists', yTitle = 'Number of Minutes', xTitle = 'Artists', width = 2)

#Making a plot showing Artist Wise Streaming Time in Hours
songs_by_artists.iplot(color='red', x= 'Artist', y = 'Hours Played',
title='Hours Played and Artists', yTitle = 'Number of Hours', xTitle = 'Artists', width = 2)

#Making a plot showing Artist Wise Streaming Time in percent
songs_by_artists.iplot(color='red', x= 'Artist', y = 'Percent of Overall Time (in %)',
title='Percentage Played and Artists', yTitle = 'Percent of Overall Time (in %)', xTitle = 'Artists', width = 2)

#Pie chart is for only those artist who have Percent of Overall Time (in %) greater than 0.1%
import plotly.express as px
df = songs_by_artists[songs_by_artists['Percent of Overall Time (in %)']>0.1]
fig = px.pie(df, values='Hours Played', names='Artist', title='Pie Chart of Favourite Authors')
fig.show()

Result

A. Taylor Swift is my favourite artist. (Why am I not surprised!😂)
B. What is surprising is that I've listened to Taylor 47.8% of time!
C. My top 10 artists are:

1. Taylor Swift
2. Céline Dion
3. Ed Sheeran
4. Shawn Mendes
5. Alan Walker
6. Lady Gaga
7. Maren Morris
8. Camila Cabello
9. Dua Lipa
10. Christina Perri

5. Song Wise

#Rounding off to two digits after decimal
songs_by_time['Minutes Played'] = songs_by_time['Minutes Played'].apply(lambda x:np.round(x,2)) 

#Cleaninng the song name
songs_by_time['Song'] = songs_by_time['Song'].apply(lambda x:x.split('(')[0])
songs_by_time['Song'] = songs_by_time['Song'].apply(lambda x:x.split('-')[0])

Plotting the top n songs

#Making a plot showing songs Wise Streaming Time in Minutes
def plot_n_songs(n):
    songs_by_time[0:n].iplot(color='green', x= 'Song', y = 'Minutes Played',
    title=f'Top {n} Songs ', yTitle = 'Number of Minutes', xTitle = 'Song', width = 2)
# plt.xticks(rotation=90)

#Top 10
plot_n_songs(10)

#Top 20
plot_n_songs(20)

#Top 50
plot_n_songs(50)

import plotly.express as px
def pie_for_n(n):
    df = songs_by_time[0:n]
    fig = px.pie(df, values='Minutes Played', names='Song', title=f'Pie Chart of Favourite {n} Songs')
    fig.show()

# Pie Chart for favourite 10 songs
pie_for_n(10)

# Pie Chart for favourite 20 songs
pie_for_n(20)

# Pie Chart for favourite 50 songs
pie_for_n(50)

#Top 50 Songs
top_fifty = songs_by_time[0:50]
#Adding Hour Column
top_fifty['Hours Played'] = np.round(top_fifty['Minutes Played']/60,2)

top_fifty[0:20]

	Song	Minutes Played	Hours Played
0	willow	1257.92	20.97
1	evermore	1190.79	19.85
2	exile	1124.17	18.74
3	august	1068.40	17.81
4	Enchanted	1019.12	16.99
5	You All Over Me	995.28	16.59
6	Let's Talk About Love	901.29	15.02
7	Untouchable	856.45	14.27
8	Afterglow	832.94	13.88
9	Delicate	831.24	13.85
10	champagne problems	827.60	13.79
11	Perfect	790.90	13.18
12	Shape of You	785.98	13.10
13	The Bones	745.70	12.43
14	Where Were You In The Morning?	694.03	11.57
15	Darkside	683.20	11.39
16	A Thousand Years	666.50	11.11
17	I'm Alive	631.53	10.53
18	All Too Well	630.92	10.52
19	My Heart Will Go On	619.63	10.33

Result

A. willow is my most streamed song. Not surprising!
B. Now comes the interseting fact, willow has been streamed for 1258 minutes. That is 21 Hours!
C. Top 10 songs are:

1. willow
2. evermore
3. exile
4. august
5. Enchanted
6. You All Over Me
7. Let's Talk About Love
8. Untouchable
9. Afterglow
10. Delicate

6. Number of Times a song is played?

#Creating a DataFrame containing the number of times a song has been played
df = songs_raw.groupby('Song').count().sort_values(by='Date', ascending=False)
songs_played = df.drop(['Artist', 'Date'], axis =1 )
songs_played=songs_played.reset_index()
songs_played.columns = ['Song', 'Number of Times']
songs_played = songs_played[songs_played['Number of Times']>2]

#Cleaning the Song Name
songs_played['Song'] = songs_played['Song'].apply(lambda x:x.split('(')[0])
songs_played['Song'] = songs_played['Song'].apply(lambda x:x.split('-')[0])

songs_played[songs_played['Number of Times']>5]

	Song	Number of Times
0	willow	408
1	You All Over Me	319
2	Afterglow	316
3	august	287
4	exile	287
...	...	...
131	Stitches	6
132	Breathe	6
133	I'll Never Love	6
134	The A Team	6
135	Dancing With Our Hands Tied	6

136 rows × 2 columns

songs_played[songs_played['Number of Times']>5].iplot(kind = 'bar', color='violet', y= 'Number of Times', x='Song',
title='Song and Times Played', yTitle = 'Songs', xTitle = 'Number of Time Played')

#A dictionary containing number of songs played at least a number of times
num_songs = {}

for n in range(0,350,10):
    if n==0:
        num_songs[5] = len(songs_played[songs_played['Number of Times']>5])
    else:
        number = len(songs_played[songs_played['Number of Times']>n])
        num_songs[n] = number
# for n in range(50,400,50):
#     number = len(songs_played[songs_played['Number of Times']>n])
#     num_songs[n] = number

#Creating a DataFrame for the same
songs_number = pd.DataFrame(num_songs, index = num_songs.keys())[0:1].transpose()
songs_number = songs_number.reset_index()
songs_number.columns =['Maximum Number of Time Played', 'Numbers of Songs']

# songs_number

#Making a barplot showing number of times a song is played with number of songs
songs_number.iplot(kind = 'bar', color='blue', x= 'Maximum Number of Time Played', y='Numbers of Songs',
title='Maximum Number of Times a Song is Played', yTitle = 'Numbers of Songs', xTitle = 'Maximum Number of Time Played')

df = songs_number[songs_number['Numbers of Songs']>1]
fig = px.pie(df, names='Maximum Number of Time Played', values='Numbers of Songs', title='Number of Songs Played a Maximum Given Times')
fig.show()

songs_played.head(10)

	Song	Number of Times
0	willow	408
1	You All Over Me	319
2	Afterglow	316
3	august	287
4	exile	287
5	evermore	279
6	Shape of You	260
7	The Bones	246
8	champagne problems	244
9	Darkside	241

Results

A. willow is the most streamed song which is streamed 400+ times!
B. There are 76 songs streamed over 50 times.
C. 55 songs streamed over 100 times.
D. 36 songs streamed over 150 times.
E. 16 songs streamed over 200 times.
F. 7 songs streamed over 250 times.
G. 3 songs streamed over 300 times.

7. Number of Songs per Day?

#Not ignoring any songs
songs_raw.groupby('Date').count()['Minutes Played'].mean()

52.316602316602314

#Ignoring songs which has been played less than 1.5 minutes
songs_g1 = songs_raw[songs_raw['Minutes Played']>1.5]

songs_raw.shape[0] - songs_g1.shape[0]

2811

average = songs_g1.groupby('Date').count()['Minutes Played'].mean()

print(f"Average number of songs is {np.int(average)}!")

Average number of songs is 41!

Result

Average number of songs is 41!