Web Scraping L'oreal: Find the One you really want

Posted on Feb 3, 2020


As a beauty shopper myself, I have a bunch of experience in selecting the most suitable product for myself. Unlike other products that have similar functionality to all kinds of consumers, there are lots of factors that consumers need to consider when selecting beauty products, e.g. Skin Type, Undertone, etc. In other words,  in order to select the one that is most suitable for them, consumers have to spend a lot of time investigating those products.

However, not every single person would be that patient and therefore sometimes consumers would simply go with the one that has high ratings, as they believe high rating products would be suitable for most people. A high rating score might indicate that most of the consumers who have purchased this product are satisfied with its quality and functionality, but does that mean it would also be suitable for you?

In this project, I am going to scrape the reviews for a specific product with high rating scores on lorealparisusa.com and let data tell me whether that particular product would be suitable for me or not by comparing good and bad reviews.



Founded in 1919, L'oreal is a French personal care company headquartered in Clichy. It is the world's largest cosmetic company and has developed activities in the field concentrating on hair-color, skin-care, makeup and perfumes.

On lorealparisusa.com, products are displayed by categories such as Skin Care, Makeup and Hair Care. For each of the product, reviews are displayed on the bottom of each product page and each of the reviews does not only contains the rating score and the review content but also the reviewer's characteristics such as eye color and undertone. Figure 1 below is an example of how reviews are displayed on the website.



In this project, I used two datasets: Products Data and Reviews Data. Both of the datasets are obtained by using Web Scraping. 

Products Data

The Products Data contains all the Skin Care, Makeup and Hair Care products on lorealparisusa.com. Each row of the dataset represents a specific product and the features are:

  • Main Category (Skin Care/Makeup/Hair Care)
  • Category (Facial Moisturizer/Face Mask/...)
  • Product Name
  • Number of Reviews
  • Price
  • Rating

Reviews Data

By using Products Data, I found the top products in each category based on their ratings. And then I scraped the reviews for Matte Lip Stain which is one of the top products. In the Reviews Data, each row represents a single piece of review and the features are:

  • User Name
  • Region
  • Age
  • Gender
  • Hair Color
  • Complexion
  • Undertone
  • Eye Color
  • Rating Score
  • Review Time
  • Title
  • Content
  • Puchased Again (Yes/No)
  • Recommend (Yes/No)



Web Scraping

For Products Data, I used Scrapy to scrape the product information. Since Scrapy does not work for the Reviews section, I used Selenium to scrapy the review information.

Data Cleaning

I used pandas and numpy for data cleaning. The tasks involved are: handling missing values, converting data types and handling strings.

Data Visualization

In this project, I used density plot, donut chart and horizental bar chart to tell the "story" behind data. Python's Matplotlib and Seaborn are used to accomplish those tasks.



Product Data

In order to get a sense of the marketing position of L'oreal, I plotted the price distribution of all the products and compared it among the three categories. 

From the plot, we can see that the price distribution for Hair Care and Makeup are similar, while the price for Skin Care products has a wider range. One possible reason could be that the volume of Skin Care products varies widely which might have a siginificant influence on prices. 

Although the price distribution in the three categories are quite different, overall the price for L'oreal products are below $40 which indicates that L'oreal is definitely not a luxury brand.

However, providing low-price products does not necessarily mean that L'oreal is providing low-quality products. If we look at the rating distribution, no matter in which category, all the ratings are centralized between 4.0 and 5.0.

By taking advantages of the rating information, I found the top products in each category and the results are shown below.

Category Product
Skin Care Bright Reveal Brightening Daily Scrub Cleanser
Skin Care Exfoliate & Refining Face Mask
Skin Care 1.5% Pure Hyaluronic Acid Serum
Makeup Lash Primer
Makeup Matte Lip Stain
Makeup Total Cover Foundation
Hair Care Extraordinary Oil Conditioner
Hair Care Extraordinary Oil Shampoo
Hair Care Extraordinary Oil-In-Cream


For this project, I decided to focus on Matte Lip Stain and therefore I scraped all the reviews for this specific product.


Reviews Data

The Reviews Data does not only contain rating information but also the reviewers' characteristics such as Age Group, Hair Coloe, etc. In order to find out the group of consumers that is most suitable for this product, I compared the good reviews and bad reviews in terms of the following characteristics. Here good reviews is defined as reviews with rating score higher than or equal to 4.0, and bad reviews is defined as reviews with rating score lower than 4.0.

  • Age Group

In the market of beauty products, many of the products has a specific age group to target. That is to say, age can have an impact on the product's functionality and therefore influence customers' satisfaction. The two figures below compare the percentage of different age groups in good and bad reviews.

From the two figures above, we can see that in good reviews around 30% of the consumers are between 18-24 years old, while in bad reviews this group of people only takes around 20%, which indicates this product might be more suitable for people aging from 18-24. Also, for people from 25-34, they might need to be more careful when making the purchase decision because the percentage in bad reviews is 8% higher than that in good reviews.

  • Undertone

Undertone is a significant factor to be considered when selecting makeup products. People with different undertones can have significantly different effect when using the same makeup product. The two figures below compare the percentage of different undertones in good and bad reviews.

From the two figures above, we can see that the percentage of neutral undertone in good reviews is 6.6% higher than that in bad reviews, indicating the matte lip stain might be more suitable for consumers with neutral undertone. On the other hand, the percentage of warm undertone in good reviews is 8.7% lower than that in bad reviews. Therefore, consumers with warm undertone might need to consider more carefully.

  • Hair Color

Similar as undertones, when using the same makeup product, people with different hair colors may look different. The figure below compares the percentage of different hair colors in good and bad reviews.

As the result indicate, I would suggest people with dark brown, medium brown or black hair to purchase this lip stain, since the percentage in good reviews is higher. However, people with light blonde, darkest brown or dark blonde hair might not be suitable for this product, as indicated by the result shown in the figure above.

  • Eye Color

Eye color also has an effect on the presentation of the product. The figure below compares the percentage of different eye colors in good and bad reviews.

As shown in the result above, the data suggests people with dark/light brown eyes to purchase the lip stain, while people with blue eyes are not encouraged to try this product.

  • Word Cloud

Last but not least, I also made a word cloud based on all the review content in order to give consumers a sense of what the key features are. Some common words such as Lip and Lipstick are taken out. 

It is obvious that many reviewers mentioned dry as a key feature,  so be careful if you are sensitive about this feature. But we could also notice other words such as long lasting and lightweight, so do not hasitate if you are looking for these features.


Conclusion & Future Work

In conclusion, the Matte Lip Stain by L'oreal would be more suitable for people who are:

  • 18-24 years old
  • Neutral Undertone
  • Dark Brown / Medium Brown / Black Hair
  • Brown Eyes
  • Looking for Long Lasting & Lightweight Lip Stains

In addition, consumers who are looking for moist lip stains are not suggested to consider this product.

Even though this project only focus on a specific product, this kind of analysis can be applied to any beauty product. Whenever you are unsure about the product you are interested in, data will tell you whether that is what you are looking for or not.

For more infomation about this project, please check out this Github Repositary



About Author


Xuyuan (Alice) Zhang

Alice recently completed her Master in Business Analytics at Brandeis International Business School. She has an interest in applying data science in the e-commerce and tourism areas. She is passionate about revealing and telling the “story” behind certain...
View all posts by Xuyuan (Alice) Zhang >

Related Articles

Leave a Comment

No comments found.

View Posts by Categories

Our Recent Popular Posts

View Posts by Tags

#python #trainwithnycdsa 2019 airbnb Alex Baransky alumni Alumni Interview Alumni Reviews Alumni Spotlight alumni story Alumnus API Application artist aws beautiful soup Best Bootcamp Best Data Science 2019 Best Data Science Bootcamp Best Data Science Bootcamp 2020 Best Ranked Big Data Book Launch Book-Signing bootcamp Bootcamp Alumni Bootcamp Prep Bundles California Cancer Research capstone Career Career Day citibike clustering Coding Course Demo Course Report D3.js data Data Analyst data science Data Science Academy Data Science Bootcamp Data science jobs Data Science Reviews Data Scientist Data Scientist Jobs data visualization Deep Learning Demo Day Discount dplyr employer networking feature engineering Finance Financial Data Science Flask gbm Get Hired ggplot2 googleVis Hadoop higgs boson Hiring hiring partner events Hiring Partners Industry Experts Instructor Blog Instructor Interview Job Job Placement Jobs Jon Krohn JP Morgan Chase Kaggle Kickstarter lasso regression Lead Data Scienctist Lead Data Scientist leaflet linear regression Logistic Regression machine learning Maps matplotlib Medical Research Meet the team meetup Networking neural network Neural networks New Courses nlp NYC NYC Data Science nyc data science academy NYC Open Data NYCDSA NYCDSA Alumni Online Online Bootcamp Online Training Open Data painter pandas Part-time Portfolio Development prediction Prework Programming PwC python python machine learning python scrapy python web scraping python webscraping Python Workshop R R language R Programming R Shiny r studio R Visualization R Workshop R-bloggers random forest Ranking recommendation recommendation system regression Remote remote data science bootcamp Scrapy scrapy visualization seaborn Selenium sentiment analysis Shiny Shiny Dashboard Spark Special Special Summer Sports statistics streaming Student Interview Student Showcase SVM Switchup Tableau team TensorFlow Testimonial tf-idf Top Data Science Bootcamp twitter visualization web scraping Weekend Course What to expect word cloud word2vec XGBoost yelp