MayaKitzis/sephora_products

🧴 Sephora Products Dataset — EDA & Insights

Sephora is a global beauty product store that sales products from thousands of brands. The dataset I will present in the project includes 8,494 records represent a different product (some from the same company) and 27 features such as product name, price, size, rating, category. This project analyzes Sephora’s product dataset describing product attributes, pricing and popularity indicators. The goal of the analysis is to explore whitch factors such as price or category, most influence product popularity and customer engagment on Sephora's playform.

The target variable in this dataset is "love_count", whitch represent the number of users who marked a product as lover or favored. It serves as an indicator of product popularity and the goal is to predict this value based on other product features such as price, rating, brand and category.

In order to acieve the most accurate conclusions I cleaned the data befoure analyzing it. I deleted duplicated data, replaced incorect prices like 0 or negative number to the median price so the incorect number won't affect on my results. In addition I searched to 99% value in the price column and deleted all the prices above it, so it won't affect on the results. I created few research questions to get a better understanding on the relations between the features whitch I present here with visualization.

---

Research Questions & Visual Insights

1. Which product categories are the most popular?

Insight: Makeup and Skincare categories dominate product popularity, while niche segments like Mini Size or Tools perform less strongly. It is a very important insight to understand which group of products we will mostly see in the other reaserch questions, which analyze all the products as one without separating between the grpoups of products.

---

2. Does product rating affect popularity?

Insight: Higher-rated products tend to receive more “loves,” suggesting a positive correlation between customer satisfaction and popularity. It can offer the lead to another question - how many people buy a product just becouse it has a high rating? does the people we see here that loved a product with high rating loved it just becouse this is the only one they used? We can also see here products with lower rating that people loved.

---

3. How does price affect product popularity?

Insight: It is very clear that price play a big role in the popularity of a product. Each product has a wide range of pricing, so people know that the expensive ones can always be replaced with a cheaper one. Mid-priced products tend to perform best while very expensive products are less popular. We can also see that very cheap ones are less rated. From this research question we can see clearly that althogh the low price, the product is still very loved, that means that the quality is not depands diractly on the price.

---

4. Are “new” or “limited edition” products more loved?

Insight: New and limited addition are less attractive that regular products but “New” products attract slightly more attention, while limited editions have mixed popularity, likely depending on brand and hype. People most likely understand that if a product is here for a limmited time, they can miss it. It creates a FOMO on part of the people and it makes me wonder on the rating of the "Limited Addition" products becouse people buy them mostly becouse they afraid to miss something that is available only for a limited time.

---

5. Which brands are the most loved?

Insight: A few luxury and cult brands consistently dominate, receiving high average engagement (“loves”) across all their products. People love buing from popular brands. It give them security in their choice knowing many other uses the products from the compay they just bought. But a popular brand is not equal to a high rating brand.

---

Additional Findings

Insight:
It is cleare th at there’s a moderate correlation between number of reviews, ratings, and popularity (loves_count),
showing that user engagement metrics tend to move together. Each indicator affects the others and togater they show a whoe picture on the products.

---

Summary & Conclusions

• Ratings and reviews strongly influence product popularity.
• Mid-range prices achieve the highest engagement.
• New releases spark user interest.
• Skincare and Makeup dominate the market by popularity.

Reflection:

This projects was a good experiance for me. I chose a topic that I already have knowladge at and it made my work much more interesting. I had a hard time choosing my research questions because each question led me to another one. I finally chose to write my forword questions that came up in the analysis in the insights of each research question and keep my research questions more general on the data instead of focusing on certain areas in the data.

Video Link:

https://youtu.be/Tt5RUuZIeTw