Spotify

It is no surprise, that Spotify’s active listening numbers are on the rise in markets hardest hit with Covid-19 and lockdown orders, as people are staying home more and have more time to listen to music. As people use the platform more, they will want new songs to listen to sooner or later, as they get bored.

That is why we propose using the current Spotify recommender engine and analyzes of historic data in order to predict what kind of music will be produced in the future. So that Spotify can make the relevant content available on the platform ahead of the emerging trend.

Therefore, we need to find answers to the following questions:

● Who are the most popular artists over time?

● How is people’s taste in music changing over time?

● How is popularity of music changing over time?

● How can historic data be used to predict what music will be produced in the future?

Spotify Dataset 1921-2021

● Audio features of 175k+ songs released between 1921 and 2021

● Collection Methodology: Collected from Spotify Web API

● Find data grouped by artist, year, or genre in the data section

● Each row represents a single track, each column represents a field of the track (audio features and identifiers) 160k+ track

● Numerical:

- acousticness (Ranges from 0 to 1)

- danceability (Ranges from 0 to 1)

- energy (Ranges from 0 to 1)

- duration_ms (Integer typically ranging from 200k to 300k)

- year (Ranges from 1921 to 2021)

- instrumentalness (Ranges from 0 to 1)

- valence (Ranges from 0 to 1)

- popularity (Ranges from 0 to 100)

- tempo (Float typically ranging from 50 to 150)

- liveness (Ranges from 0 to 1)

- loudness (Float typically ranging from -60 to 0)

- speechiness (Ranges from 0 to 1)

..................................................

Let's Get Started

Load Dataset

Set working directory

Constructing data frame

Inspect dataset

the dataset shows 19 columns

The dataset shows 174,389 rows and 19 columns

With 15 numeric columns and 4 string columns

There are 172,230 unique values in "id" column

There are 137,013 unique values in "name" column

There are 36,195 unique values in "artists" column

There are 11,043 unique values in "release_date" column

Data Cleaning

Check for missing values

................................................

Explore The Most Popular Artists

The Top 20 Most Popular Artist in Dataset with More Than 100 Tracks

........................................

Correlation Test

Correlation Visualization

Questions arise from constructing correlation test

1. Why positive correlation on "popularity" and "year"?

Based on the correlation test,"popularity" postively correlates with "year", with a correlation of "0.513227"

2. Why negative correlation on "acousticness" and "year"?

Based on the correlation test,"acousticness" negatively correlates with "year", with a correlation of "-0.607515"

Understanding “Popularity”

the songs in the dataset which produced during 1950s and 2000s were more popular.

Spotify provides fewer songs before the World War II, maybe because there were not many songs were produced during that period. While Spotify carries steady number of songs produced during 1950s and 2000s. However, number of songs produced after 2000s is fluctuated. There was once Taylor Swift took her songs away from Spotify.

Acousticness

By excluding the factor "number of counts", we see a clearer trend acousticness is trending downwards, meaning that people no longer listen to acoustic songs as much as they did in the past. Electric & Rap songs are becoming the new trend Technology has been used in the songs as technology gotten better Disco & more electronic during early 1990s were becoming less popular, and people became more interested in more acoustic music.

The drop from 1950s - 1980s was steeper, the shift was more rapid.

In the early 1990s we see a slight increase in the popularity of acoustic song again, likely due to the fact that people wanted a break from the electronically aided disco music that was popularized the 80s.

2000s - present, there was more variation in the styles of music that were popular, as evident by the more spread out nature of the dots on the graph below.

Acoustic music is more and more creating its own popular niche, as can be seen the greater variety of acousticness levels being popular in recent years.

In general

1. The popularity of songs is getting higher, meaning people are rating songs higher nowadays than in the past

2. People are less likely to listen to songs with acousticness

3. People are more likely to listen to songs with higher level of energy & loudness

Analysis And Reasoning Behind The Findings

Music history:

(1) Before 1940s, the music industry wasn't that organized and complete, it was something new and untouched.

(2) This time was also before - during the war(World War II). After the war ended, people started having more leisure time to create songs.

(3) The 1950s is the beginning when the Beatles came into public eyes. The music industry started to do more productions after seeing that people started to like this type of music.

(3) Nowadays, more and more songs are being produced. People are considering music as an important part of their daily lives.

The rise of technology:

(1) Music producers started to utilize technology in the music industry. The acousticness for songs started going downwards.

(2) The market size of the music industry started to increase and the amount of investment in the industry became larger, especially on the use of technology.

Recommendations:

(1) Since people are now considering music an important part of their lives, the music industry is likely to continue growing. Spotify should continue to increase their number of music they have in their system to attract more and more listeners.

(2) People are less likely to listen to songs with acousticness. Music producers should focus more on combining music with technology. Also, Spotify should collaborate with more music creators that are creating high-energy, electric songs to receive high popularity ratings and to increase their customer base.