Skip to content

This project analyses skin conditions and recommends products availabel on Amazon for specific skin types

Notifications You must be signed in to change notification settings

OliviaHelena10/skincare_recommendations

Repository files navigation

Skincare Recommendations

skincare-101-800x445

Context

Skincare has become increasingly popular in recent years. Regardless of gender, identifying your skin type and establishing a personalized skincare routine is crucial for maintaining healthy, radiant skin. A tailored approach can significantly enhance your appearance and overall skin health.

Objectives

  • Create a AI that identifies skin types and recommends products tailored to specific skincare needs.
  • Understand which skincare products have the greatest impact on sales and why.

Datasets

The datasets used in this project were sourced from Kaggle with implementation from chatgpt to make them more complete:

Results

The initial results were not as strong as expected, even after experimenting with various models. The validation accuracy fell short of what is considered efficient. After multiple tests, it became evident that the primary issue lies within the data itself. This is something that will be addressed in future iterations. You can try out the current version of the app, where you can upload an image to determine your skin type: Online Skin Type Classifier. Once the data and code are refined, the user experience and accuracy should improve.

Pipeline:

pipeline_skins

Machine Learning Algorithm

The model architecture includes one rescaling layer, one padding layer, one reduction layer, and two hidden layers (the first with 6 filters, and the second with 128 neurons).

model_ml

Process

After downloading the necessary datasets, I began exploring the data using Visual Studio Code. I realized that I needed additional libraries that I hadn’t worked with before, so I searched for online resources and tutorials to guide me through the challenges I encountered.

For my machine learning model, I consulted various videos, articles, and documentation to deepen my understanding. I experimented with the tools and eventually published the results on a Streamlit application.

Once the model was complete, I turned my attention to the Amazon skincare products dataset to extract insights. Below is one example of the insights I discovered:

grafico

References

About

This project analyses skin conditions and recommends products availabel on Amazon for specific skin types

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published