Industry Presentation Slides
We propose to assist music producers with their search for music production opportunities. Our slices of analysis focus on four hypothesis test questions:
(1) Is there a statistically significant difference between the scores of albums in the jazz and metal genres?
(2) Will the label with the most artists have the highest album scores?
(3) Is there a relationship between labels' genre diversity and labels' average score?
(4) Are the average album scores of 2000-2009 different from those of 2010-2017??
Our methodology consists of approaching each of these questions with a null hypothesis to reject and an alternative hypothesis to accept in the case of rejection. Our hypotheses for each question are as follows:
-
(1) H0: Jazz albums have the same average score as metal albums.
-
(1) Ha: Jazz albums do not have the same average score as metal albums.
-
(2) H0: The label with the most artists signed will have the same average score as the label with the least artists signed.
-
(2) Ha: The label with the most artists signed will not have the same average score as the label with the least artists signed.
-
(3) H0: The label with the most genre diversity will not have a higher average score than the label with the least genre diversity.
-
(3) Ha: The label with the most genre diversity will have a higher average score than the label with the least genre diversity.
-
(4) H0: Average album scores are the same in both the 2000-2009 and 2010-2017 periods.
-
(4) Ha: Average album scores have changed in the more recent period.
During our statistical analysis, we found the following results:
(1) Jazz albums consistently earn scores slightly higher than metal albums. Using a two sample two-tailed z-test, with a significance level of 0.05 and a p-value of 7.5e-06, we reject our null hypothesis in favor of our alternative hypothesis. Jazz-Metal Comaparison Notebook
(2) The label with the most artists signed does not perform significantly differently than labels with only one artist signed. Using a two-sided T-test, with a significance level of 0.05, and a p-value of 0.706, we fail to reject the null hypothesis in favor of the alternative hypothesis. Artist Density Notebook
(3) The labels with the most genre diversity does have statistically significantly higher scores than the labels with the least genre diversity. Using a one sided T-test, with a significance level of 0.05, and a p-value of 0.015 we successfully reject the null hypothesis in favor of the alternative hypothesis with a statistical power of 58.7%. Genre Diversity Notebook
(4) Albums released between 2000 and 2009 consistently earn scores slightly higher than albums released between 2010 and 2017. Using a two sample two-tailed z-test with a significance level of 0.05 and a p-value of 5.67e-08, we reject our null hypothesis in favor of our alternative hypothesis. 2000s-2010s Comparison Notebook