The goal of this project was to build a predictive model to help Airbnb users in Lyon identify the optimal market price for their listings. By entering property details in a form, users received data-driven price recommendations to remain competitive.
Steps followed:
-
Data Preprocessing and EDA
Imputed specific variables using a K-Nearest Neighbors (KNN) Regressor to enhance data completeness and accuracy. -
Model Preparation
Analyzed histograms of key variables.
Tested for skewness and applied logarithmic or Box-Cox transformations as needed, followed by re-evaluating histograms.
Examined variable correlations to identify potential predictors.
Encoded categorical variables such as Superhost (True/False), Room Type, and License Type. -
Model Building
Implemented a Random Forest Regressor to build the prediction model. -
Interactive Dashboard Design
Created a Power BI dashboard to visualize insights and demonstrate how Airbnb listing prices are affected as the variables included in our prediction model are adjusted or modified. -
Result Presentation Application
Developed a Streamlit app to showcase the findings, incorporating the predictive model and an interactive, form-based interface. Users can input key property details, such as neighborhood, number of bedrooms, number of bathrooms, host type and other relevant characteristics, to generate an estimated listing price for the Airbnb based on the model's predictions.
This project requires CSV files that are not included in the repository.
- Datasets from: Inside Airbnb.
