The Sound of Ghana

RANKING METHODOLOGY

At the core of this ranking system is a comprehensive evaluation of the top-performing songs as of December 31st, 2024. The methodology involves a multi-step process that combines metrics from YouTube performance, user engagement, search trends, and sentiment analysis. Songs were tallied and scaled to a top 20 chart, with each song’s corresponding video performance playing a major role in determining its final position.

The YouTube videos—both official music videos and visualizers—were extracted and ordinally ranked based on the number of views. Notably, visualizers, which are typically animated or static visual content used before the release of official videos, gained significant traction in 2024. In some instances, visualizers surpassed official music videos in view count. When this occurred, the visualizer was considered for the ranking instead of the official video, underscoring the shifting landscape of music video consumption.

Comment Analysis

Comments associated with these videos were harvested and subjected to natural language processing (NLP) and machine learning analysis. To ensure data integrity, all comments posted after December 31, 2024, were excluded from the dataset. In preparing the data, all 'NaN' (not a number) entries found in the comment fields were replaced with empty strings to avoid errors during text-based analysis. Similarly, 'NaN' values in the 'Reply Count' column were replaced with zeros so that the dataset could be processed numerically without issue. The 'Date' column was also converted to DateTime format to aid in time-series trend analysis.

Search Interest

The methodology also included an assessment of public interest through cumulative search activity. This involved analyzing search trends on Google and related platforms, using a suggestions extension tool to refine results. For example, when users searched for a name like “Lasmid,” unrelated searches for other individuals with the same name could appear. The extension allowed the isolation of each artiste's profile by assigning a unique identifier, ensuring that the search interest was accurately attributed. To standardize the search interest values, the highest point of search volume for each artiste was set at 100, with all other data points scaled in relation to this peak.

Geographic Interest

Interest in Ghanaian artistes extended beyond the country’s borders. Analysis of regional data revealed that Liberia had the highest level of interest in Ghanaian music outside of Ghana. This was followed by Equatorial Guinea, Togo, Zambia, and Nigeria. Additionally, three artistes—DopeNation, DJ Vyrusky, and Beeztrap KOTM—showed notable international reach, gaining traction even beyond the African continent.

Figure 1: Relative search interest in artistes (Africa).

Figure 2: Percentage distribution of interest in artistes (Africa).

Sentiment Analysis

To understand audience engagement on a deeper level, a sentiment analysis was performed on a sample of user comments. These comments were randomly selected and categorized into various sentiment classes using large language models. The analysis uncovered recurring expressions of love, admiration, and enthusiasm for the artistes. Comments that did not directly reference the song or the artiste were categorized under a neutral class labeled “others.” This helped differentiate between genuine fan interactions and unrelated commentary.

Overall, the methodology is a blend of quantitative and qualitative evaluation, offering a comprehensive lens through which to interpret musical impact, digital engagement, and regional popularity.