Final Project Submission - Vu.docx

Final Project/Raw Dataset

@@ -0,0 +1 @@
+https://dickinson0-my.sharepoint.com/:x:/g/personal/phamduyd_dickinson_edu/ETCYhqxdnFFBuwQGn4aU0jIBWhzsO48ybIblqPPPgIyokg?e=BxrMQD

Final Project/ReadMe.md

@@ -0,0 +1,63 @@
+# Predicting the "Perfect" Airbnb Listing
+##### By Pham Duy Anh Vu, Hannie Pham
+
+---
+
+## Introduction
+
+When thinking of an ideal Airbnb listing, we usually think about compelling photography, clear and concise descriptions, responsive hosting, and a curated selection of amenities that cater to guest needs. It sets transparent expectations around space, price, and house rules, while highlighting unique features—be it a cozy reading nook, a stunning city view, or thoughtful local touches—that resonate with travelers and encourage bookings.
+
+This study aims to quantitatively assess these elements and uncover key factors behind high-performing Airbnb listings, defined as those with guest ratings above 4.5, offering practical guidance for hosts, guests, and platform designers aiming to enhance listing quality and user experience.
+
+---
+
+## Data & Methods
+
+- **Source:** [Airbnb.com](https://www.airbnb.com/) (via [`airbnb_api`](https://apify.com/tri_angle/airbnb-scraper))
+- **Preprocessing:** 
+  - We first scraped our data using the `airbnb api` which returned the raw data onto a csv
+  - We then used the `pandas` Python library to clean and process the listing data to prepare it for analysis
+- **Visualization / Models:**
+  - Used the `ydata_profiling` Python package for some preliminary analysis and visualizations
+  - All visualizations were created using `matplotlib` and `seaborn`
+  - We used linear regression models and Random Forest classification models for the predictive analysis
+
+---
+
+## Project Recreation
+
+- **File structure:** All of the documents related to this project lies within the 'Final Project' folder
+  - **Raw Dataset** — The raw dataset file
+  - **cleaned_airbnb.xlsx** — The cleaned dataset file
+  - **Data400 Final Project.ipynb** — The Jupyter notebook that holds the Python code that runs the modeling and analysis for the project
+  - **ReadMe.md** — The ReadMe file
+  - **Data400 - Final Presentation.pptx** — The Powerpoint which compiles the results and discussion of the project
+
+- **Recreation steps:** 
+  - **Step 1:** Download or clone the project files
+  - **Step 2:** Open the 'Data400 Final Project.ipynb' file, change the working directory to your current working directory
+  - **Step 3:** Run the codes step-by-step to reproduce the results
+
+---
+
+## Results Summary
+
+- **Listing Type and Pricing:** Boutique hotels, villas, and hotels had the highest median prices, while guesthouses and tiny houses offered more budget-friendly options with occasional high-end outliers.
+- **Amenity Insights:** Listings with popular amenities like Wi-Fi, kitchen, and air conditioning tended to receive higher ratings, suggesting guest preferences significantly impact listing success.
+- **Key Predictors of High Ratings:** Important features influencing listing success included number of reviews, amenities offered, superhost status, and listing type.
+- **Regression Model Performance:** Linear regression and random forest models were used to predict both price and rating. The random forest model achieved better performance with higher accuracy and lower error metrics.
+
+---
+
+## Implications
+
+- Understanding key drivers of high ratings can help hosts optimize listings for better guest satisfaction and higher bookings.
+- Guests can use insights about pricing, property types, and common high-rated features to make more informed booking decisions aligned with their preferences and budget.
+- The predictive models were able to find notable characteristics that affected ratings but remained inaccurate
+
+---
+
+## Future Work
+- Experiment with more models and tuning methods to generate more accurate predictions
+- Expand the analysis to include host-level data
+- Possible intergration of data such as 'local crime rate' or 'hotel prices' to deepen on location-based success