Background

The Project Report (A2) implements the methods of analysis outlined in the Project Plan (A1), and poten- tially extends the original analysis plan using some of the more advanced topics covered in the latter half of the semester. The report should be comprehensive – sketching out all aspects of the project from beginning to end. You should draw on proposal, particularly in the opening sections of your report.

The report is not expected to provide definitive answer to the business problem, rather it should provide

convincing evidence about what measures could be taken to address the problem. For example, the insights derived from your linear regression model should provide clues as to what kind of music could be selected to optimise their business objectives (e.g. increase engagement in a marketing domain, or maintain profitability in an accounting domain).

Sections of the Project Report (A2)

The project report will likely have the following sections.

1.  Background

2.  Business problem

3.  Method of Analysis

4.  Results of Analysis

5. Strategic Recommendations

The first three of these sections, the Background, Business Problem, and Method of Analysis will be based on revisions to your Project Plan (A1). It is OK to use the same text used in the Project Plan (A1), but given the key deliverable of this project is in the form of a slidedeck you should reduce/summarise these sections from the plan when integretaing them into the report.

The Results of Analysis section presents your findings from working with the Million Song Data (MSD).

The structure of this section should start with basic visualisations of the data and descriptive statistics from

the data (e.g., the number of observations used in your analysis, statistical summaries and assumption tests for variables included in your analysis) followed by the results of linear regression model(s).

The results of your analysis should be accompanied by an interpretation of each of the outputs you include in your report. These may be presented concurrently on the same slides as your results or be presented on subsequent slides – the choice is yours. All interpretations need to be correct (e.g., to what extent the inputs to your regression model predict the outcome variable, additional insights derived from extensions to linear regression). Remember, the goal in conducting this analysis is not to find a particular solution (i.e., there is no one “correct”solution), rather the goal is to attempt to generate value for the client using the Million Song Data (MSD). Null or statistically insignificant results can provide just as valuable insights as statistically significant results. The interpretations of linear regression modelling and any extensions to linear regression are likely to be the main focus of this section, but consider also that there may be highly relevant insights to derive from your visual and descriptive analytics (in some cases, these may tell a more convincing story than the more statistically sophisticated modelling).