data-portfolio

Airbnb Listings — Bangkok Market Analysis

Exploratory analysis of Bangkok Airbnb listings to understand price drivers, neighborhood patterns, and levers to improve occupancy & revenue.

---

Project Overview

• Title: Airbnb Bangkok — Price & Demand Patterns
• Business problem: Which factors (room type, neighborhood, availability, minimum nights) drive price and potential occupancy, and how can hosts/platforms use them to optimize pricing, supply, and listing strategy?
• Dataset: Airbnb Listings Bangkok.csv — 15,854 rows × 17 columns. Core fields: price, room_type, neighbourhood, latitude, longitude, minimum_nights, number_of_reviews, last_review, reviews_per_month, availability_365, calculated_host_listings_count. Date range for last_review: 2012-12-15 → 2022-12-28.
• Tools used: Python (Pandas, NumPy, Matplotlib), Jupyter Notebook.

---

Methodology (Concise)

1. Load & schema checks Remove index col (Unnamed: 0), confirm dtypes; parse last_review to datetime.
2. Cleaning Handle missing values (last_review, reviews_per_month), cap extreme outliers for visualization (e.g., price above P99).

3. Feature prep

   • Create robust statistics (median/quantiles) due to right-skewed price.
   • Aggregate by room_type and neighbourhood; derive simple demand proxy from reviews_per_month and availability_365.
4. EDA & visualization Distribution plots (price, availability, minimum nights), heatmaps/boxplots by room type & neighborhood, and basic geo overview.
5. Validation Sanity checks (top neighborhoods by count, price quantiles, outlier review of min nights & price).

---

Analysis Results

• Key finding 1 — Price drivers Prices are right-skewed (median ≈ 1,429; P95 ≈ 6,762; max outliers exist). Median price by room type:

  • Hotel room ≈ 1,700
  • Entire home/apt ≈ 1,536
  • Private room ≈ 1,212
  • Shared room ≈ 500 → Room type is a strong discriminator; “entire place” and “hotel room” command higher medians than private/shared rooms.

• Key finding 2 — Geographic concentration & pricing Listings are concentrated in Vadhana and Khlong Toei (≈ 26.8% of all listings combined). Median neighborhood prices (top examples): Nong Chok ~ 2,539, Parthum Wan ~ 2,400, Vadhana ~ 2,000, Bang Rak ~ 1,850. → Neighborhood effects are material and can inform localized pricing.

• Business impact

  • Dynamic pricing by room type & neighborhood reduces under/over-pricing risk.
  • Minimum nights: majority allow 1 night (P75 = 7) but outliers (up to 1,125) depress demand—tighten rules to market norms.
  • Supply strategy: expand in high-demand clusters (Vadhana/Khlong Toei); optimize availability for seasonality (median availability_365 ≈ 309 days; median reviews_per_month ≈ 0.44).

---

Dashboard & Files

Adjust the relative paths below to match your repo if needed.

• Notebook: Airbnb Listings Bangkok Analysis.ipynb

• Dataset (CSV): Airbnb Listings Bangkok.csv

---

Technical Highlights

• Size: 15,854 records, 17 fields; 6,659 unique hosts; 50 neighborhoods.

• Distributions:

  • price—median 1,429, mean 3,218 (skewed by outliers), P75 2,429, P95 6,762.
  • availability_365—median 309 (IQR 138–360).
  • minimum_nights—median 1 (P75 7), max 1,125.
• Room-type medians: Hotel 1,700 > Entire 1,536 > Private 1,212 > Shared 500.
• Neighborhood concentration: Vadhana (2,153 listings), Khlong Toei (2,097 listings).

---

Skills Demonstrated

Data wrangling (Pandas), exploratory data analysis, outlier handling, robust statistics, visualization, and business interpretation for pricing & supply decisions.

---