You are on page 1of 3

1) Can we predict the popularity of a song given the values of the other variables available in

the dataset?

The above image is the output of the multiple linear regression. We can predict the popularity of
a song given the values of other variables in the dataset. We have to put coefficients near
predictors and get predictor values from the given dataset, thus we would be able to get the
popularity of the song.

2) Given the other variables, is danceability significantly related to popularity?


Looking at the same image above. We can see the estimate for danceability estimate is -20.49
and its p-value is 0.0708 which statistically significant at the level of 0.1. We can the significant
code from the output, near the danceability estimate there is a dot, we can see that it
corresponds for the significance level of 0.1. Thus, we can conclude that danceability is
significantly related to popularity.

3) If you wish to author a popular song on Spotify, which features would you mostly consider?
Again, from the image above which is the output of linear regression we can see that:
danceability
energy
mode
time_signature
These variables are significantly related to popularity. It might be important to consider those
features mostly. Then from correlation plot, we would see which other predictors are closely
correlated with popularity of a song.

MRL diagnostics:

You might also like