Getting started

Click here to open the slides full screen.

from setup import ___

from siuba import *
from plotnine import *

from music_top200 import music_top200, track_features

Exercise 1:

In this exercise, there are two code cells. The first defines variables for tracks by different artists. The second creates a plot.

Read through the code and plot, and then modify it to answer the question beneath.

roddy = (
  track_features
  >> filter(_.artist == "Roddy Ricch")
)

billie = (
  track_features
  >> filter(_.artist == "Billie Eilish")
)

other = (
  track_features
  >> filter(_.artist == "ITZY")
)

other
artist album track_name energy valence danceability speechiness acousticness popularity duration
1431 ITZY IT'z Different 달라달라 (DALLA DALLA) 0.853 0.713 0.790 0.0665 0.00116 73 199.874
21148 ITZY IT'z Different 달라달라 DALLA DALLA 0.853 0.713 0.790 0.0665 0.00116 57 199.874
22388 ITZY It'z Me WANNABE 0.911 0.640 0.809 0.0617 0.00795 81 191.242
25287 ITZY IT'z ICY ICY 0.904 0.814 0.801 0.0834 0.03240 72 191.142

4 rows × 10 columns

The code below plots hits for the roddy variable. Note that you could swap out roddy for any of the other two variables above.

(roddy
  >> ggplot(aes("danceability", "speechiness"))
   + geom_point()
   + expand_limits(x = [0, 1], y = [0, 1])
)

Who has the widest range of danceability? (i.e. biggist difference between highest and lowest)

Roddy Ricch Try again.
Billie Eilish That’s right!
ITZY Try again. All the ITZY songs shown have roughly the same danceability. 
# Note that this code use a special pandas function to put all
# the data together into a single variable (all_together), to
# make it easy to see all the data together.
# To answer the question, you should replace all_together, with
# the relevant variable (billie, other, etc..)

import pandas as pd
all_together = pd.concat([roddy, billie, other])

(all_together
  >> ggplot(aes("danceability", "speechiness", color = "artist"))
   + geom_point()
   + expand_limits(x = [0, 1], y = [0, 1])
)

Exercise 2:

Does it look like there any extremely popular songs over 15 minutes long?

There is not one concrete answer to this question. Make a plot below, and come up with an answer you might share with another person.

hint

The duration column contains the length of each song in seconds. Use this with the popularity column.

(track_features
  >> ggplot()

)

(track_features
  >> ggplot(aes("duration", "popularity"))
   + geom_point()
)

possible answers

screencast

Exercise 3:

Does the lowest energy track belong to a “low energy” artist? In this exercise, we’ll explore the questions using tracks by two artists.

Here is the track data sorted by energy.

# this code is to help you examine track energy levels 
(
  track_features
  >> arrange(_.energy)
)
artist album track_name energy valence danceability speechiness acousticness popularity duration
1003 Simon Smith Loops Blagaslavlaju vas 0.000778 0.000 0.779 0.4210 0.99400 0 36.038
5995 DMS Prepáčte Nič 0.000791 0.000 0.571 0.4460 0.95000 25 37.355
16689 Peter Simon Snowrain Snowrain 0.003480 0.373 0.472 0.0517 0.99600 0 31.000
... ... ... ... ... ... ... ... ... ... ...
22695 Nino Xypolitas Epireastika Eime Enas Allos - Original 0.996000 0.517 0.644 0.1030 0.00346 34 214.693
17072 Otira Soundboy Burnin’ Soundboy Burnin’ 0.997000 0.327 0.568 0.2330 0.00299 14 173.846
11069 Scooter No Time To Chill How Much Is the Fish? 0.999000 0.615 0.533 0.0786 0.00130 48 226.200

25321 rows × 10 columns

Notice that Simon Smith has the lowest energy song (“Blagaslavlaju vas”), while Scooter has the highest energy song (“How Much is the Fish?”).

First, filter the track_features data to create a variable named artist_low that has only tracks by the artist Simon Smith.

# create artist_low variable here

artist_low = (
    
)

Next, create a variable named artist_high with tracks by the artist Scooter, who has the highest energy song.

# create artist_high variable here

Based on separate plots of their data, does the artist with the lowest energy track seem to have lower energy songs in general?

# examine plots of each artist, to answer question
# note, we're examining energy, so to make a scatterplot you could set
# the second variable to anything interesting (e.g. popularity)

(___

)
⚠️: Don't forget to replace all the blanks! 
artist_low = track_features >> filter(_.artist == "Simon Smith")

artist_high = track_features >> filter(_.artist == "Scooter")

ggplot(artist_low, aes("energy", "popularity")) + geom_point()

#ggplot(artist_high, aes("energy", "popularity")) + geom_point()

possible answer

The high energy artist, Scooter, seems to only have high energy songs (from about .9 to 1 energy).

On the other hand, the low energy artist, Simon Smith, seems to have a wide range of energy values (from about 0 to 1 energy).