Sentiment Analysis

*GitHub repository available here.

There are a lot of ads out there to publicize products which mostly end up not being watched or talked about by people. This ad had the opposite effect, it became a viral topic and got the attention of many, me included. Watching an ad to the end is common, what is not common is re-watching to reminisce the experience all over again.

Hence, I decided to scrape and analyze Samsung Galaxy Z Flip4 ad comments on YouTube. I scrapped the comment section to understand the opinion expressed by each individual, their personal opinion or judgment of the ad.

Outline

• Overview

• Data Collection

• Methodology

• Findings

• Conclusion

• Recommendation

Overview

This project assessed the Samsung Galaxy Z Flip4 video on YouTube with 15M views and 1600+ comments. Samsung, is a Fortune global top company recognized globally as an industry leader in technology. For this project, I delved into sentiment analysis to understand how viewers perceive the video.

Data Collection

The dataset used for this project was scrapped from YouTube. You can access the codes used for scraping here.

Methodology

This project was divided into two phases based on my desired outcomes. I treated the two phases separately because they both require different methods. There is no one-size-fits all approach hence, I had to take two approaches to get the desired outcomes.

• Phase 1: with emojis

It is optional to work with emojis depending on the use case but I believe it gives additional meaning about a text. To work with the emojis, I first had to convert the comments into a long string. Then I applied the emoji conversion function to convert the emojis to words as well as the regex and tokenize function to remove punctuations and stop words.

I ensured the emojis were converted first before applying the regex and tokenize function because doing this the other way around would remove all emojis associated with each comment. To get the stemming function working when applied to change each word into their root words, I first had to convert the string to list. After which I visualized the top words.

• Phase 2: without emojis

Here, I worked without the emojis. I converted the comments into usable formats by removing duplicates, stop words, and punctuations. Each comment was then stemmed to its root word. I conducted text polarity here to measure the positive or negative emotions in a sentence.

def polarity(Comment):
    return TextBlob(Comment).sentiment.polarity

This was done to understand the opinion expressed by each individual, their personal opinion or judgment of the ad. After which I visualized the text in charts and word cloud to get the top words.

Findings

The pie chart shows the distribution of sentiments. The ad which has 15M views and above 1600 comments received a lot of positive comments from viewers with approximately 48% of individuals dropping a positive comment.

It also has neutral comments of approximately 43%. With the high percentage of positive comments, the negatives are negligible.

The table shows the top 5 positive comments with the highest polarity measure.

Overall, the top words are nothing short of positive. Viewers dropped a lot of adjectives in the comment section including joy, love, good, laugh, and like to name a few. This indicates that a lot of people just like me finds the ad funny and entertaining with more positive and neutral comments than negative.

Conclusion

Samsung has a knack for putting out mind blowing ads and after watching the Samsung Galaxy Z Flip4 ad on YouTube I couldn't agree more and was intrigued by how entertaining and innovative the ad is. The marketing team did a great job again and judging from my analysis, 90.8% of viewers that dropped a comment share similar opinion.

Recommendation

There are tons of ads getting ignored or easily skipped out of viewers faces in seconds in all social platforms including YouTube. But unlike other apps one good thing that comes with YouTube ads for marketing is that most ads don't skip which forces users to stay glued.

This technique when utilized together with an effective ad gives the product the needed publicity. Marketing teams should take this advantage, utilize YouTube ads, and ensure they put out effective ads which attract consumers' attention, hold their interest to the message being passed, arouse their desire for the product and hopefully obtain an action.