Menu

Statistical Analysis

After identifying the collaborations profiles present in the network, we verify if there is a correlation between an artist's success and his/her collaboration behavior. This analysis took place in three steps:

Normality Test

We use the normality test of Shapiro-Wilk to verify is the data follow a normal distribution

Correlation Test

We apply the correlation test of Pearson, Spearman and Kendall and compute the t-test

Box Plot

We plot box plots of each cluster in relation to the success metrics

Normality Test

We apply the Shapiro-Wilk test with the null hypothesis that the distributions are Normal.
All p-values are much smaller than the significance level of α = 0.05

Sample Statistic p-value Normal
Popularity 0.93788 5.269e-11
Followers 0.6737 2.2e-16
Degree 0.78467 2.2e-16
Weighted Degree 0.76878 2.2e-16
eccentricity 0.74696 2.2e-16
Closeness 0.78364 2.2e-16
Betweeness 0.58553 2.2e-16
Eigencentrality 0.66979 2.2e-16
Clustering 0.87115 2.2e-16

Correlation Test

We apply the statistical correlation evaluation considering the null hypothesis that there is a correlation between the variables.

Popularity of Spotify vs. Topological Metrics

Metric Pearson Spearman Kendall Correlated
Degree 0.51 0.51 0.38
Weighted Degree 0.48 0.46 0.33
eccentricity 0.31 0.21 0.16
Closeness 0.33 0.47 0.34
Betweeness 0.44 0.51 0.38
Eigencentrality 0.43 0.49 0.35
Clustering 0.04 0.11 0.07

Number of Followers in Spotify vs. Topological Metrics

Metric Pearson Spearman Kendall Correlated
Degree 0.20 0.18 0.13
Weighted Degree 0.25 0.31 0.22
eccentricity 0.10 0.21 0.02
Closeness 0.14 0.16 0.11
Betweeness 0.19 0.21 0.15
Eigencentrality 0.16 0.15 0.10
Clustering 0.04 0.07 0.05

Correlation Matrix


Box Plot

The box plots show an analysis for each cluster related to the success metrics (popularity and number of followers).

Box Plot of popularity per cluster

Box Plot of followers per cluster