omershahar/housing-market-resilience-audit

---
license: apache-2.0
---

# Forensic Housing Market Resilience Analysis (REmatch)

<video src="https://huggingface.co/datasets/omershahar/housing-market-resilience-audit/resolve/main/Video%20Project.mp4" controls="controls" style="max-width: 720px;"></video>
## Overview
This project presents a forensic Exploratory Data Analysis (EDA) and predictive framework for the U.S. residential real estate market (2012–2023). Using the **REmatch** model, we analyze supply-side volatility to distinguish between "False Oversupply Warnings" (entry opportunities) and "Real Market Deterioration".

## Objectives
* Identify structural supply shocks vs. seasonal noise.
* Quantify market resilience using the **Three Pillars**: Behavior, Infrastructure, and Socioeconomics.
* Validate the "Luxury Veil" theory and its impact on price perception.
* Predict 10-month forward price appreciation outcomes.

## Dataset Description
The analysis utilizes the **HouseTS** dataset, covering U.S. ZIP codes with key features such as:
* `inventory_sa`: STL-adjusted inventory levels.
* `median_sale_price`: Median transaction price (Robust to outliers).
* `DRS` (Dynamic Resilience Score): Physical infrastructure strength.
* `SCS` (Seller Capitulation Score): Behavioral indicator of seller bargaining power.
* `RTI` (Rent-to-Income): Socioeconomic affordability metric.

## Data Cleaning & Pre-processing
* **Missing Value Audit:** Mapped data gaps to ensure "Investment Grade" reliability.
* **STL Decomposition:** Removed seasonal "Spring Fever" noise from inventory growth.
* **Outlier Management:** Implemented **Winsorized Medians** to negate the "Luxury Veil" distortion caused by extreme high-end sales.

## Exploratory Data Analysis (EDA) Insights
* **The Luxury Veil:** We proved that arithmetic means overstate returns by over **115%**. By switching to **Robust Medians**, we identified the true core market appreciation of ~16%.
* **Signal vs. Noise:** Using STL decomposition, we revealed that many "oversupply alarms" are seasonal artifacts. True alarms only trigger when the structural trend deviates significantly from the 12-month average.
* **Behavioral Pricing Power:** We identified a correlation where rising **Seller Capitulation Scores (SCS)** lead to a collapse in bidding wars, signaling a loss of seller negotiating power.

## Key Forensic Findings
* **Market Fortresses:** ZIP codes with high infrastructure (DRS) clear excess inventory faster, validating the "Moat" theory.
* **False Warnings:** The model identified entry opportunities where prices rose 10 months later despite initial supply alarms.
* **Critical Vulnerability:** The forensic audit highlighted potential **Survivorship Bias** in the historical data.

--- 许可证：Apache-2.0
---
# 法医视角住房市场韧性分析（REmatch）

<video src="https://huggingface.co/datasets/omershahar/housing-market-resilience-audit/resolve/main/Video%20Project.mp4" controls="controls" style="max-width: 720px;"></video>

## 项目概述
本项目针对2012年至2023年的美国住宅房地产市场，构建了一套法医视角的探索性数据分析（Exploratory Data Analysis，EDA）与预测框架。本研究借助**REmatch**模型，对供给端波动展开分析，以区分“虚假供应过剩预警”（入市机遇）与“真实市场恶化”两类场景。

## 研究目标
* 识别结构性供给冲击与季节性噪声。
* 基于“三大支柱”——行为、基础设施与社会经济——量化市场韧性。
* 验证“奢华面纱（Luxury Veil）”理论及其对价格认知的影响。
* 预测10个月后的远期房价涨幅。

## 数据集概况
本分析采用**HouseTS**数据集，覆盖美国各邮政编码区域，其核心特征包括：
* `inventory_sa`：经STL调整后的库存水平。
* `median_sale_price`：中位数交易价格（对异常值具有鲁棒性）。
* `DRS`（动态韧性评分，Dynamic Resilience Score）：实体基础设施实力。
* `SCS`（卖家让步评分，Seller Capitulation Score）：衡量卖家议价能力的行为指标。
* `RTI`（租金收入比，Rent-to-Income）：社会经济可负担性度量指标。

## 数据清洗与预处理
* **缺失值审核**：对数据缺口进行映射处理，以确保达到“投资级”可靠性标准。
* **STL分解**：从库存增长数据中移除季节性“春季热潮”噪声。
* **异常值处理**：采用**缩尾中位数（Winsorized Medians）**，以抵消极端高端销售引发的“奢华面纱（Luxury Veil）”扭曲效应。

## 探索性数据分析（EDA）核心发现
* **奢华面纱（Luxury Veil）效应**：本研究证实，算术平均值会将收益高估115%以上。通过改用**鲁棒中位数（Robust Medians）**，我们识别出约16%的真实核心市场涨幅。
* **信号与噪声**：借助STL分解，我们发现大量“供应过剩警报”实为季节性假象。仅当结构性趋势与12个月均值出现显著偏离时，才会触发真实警报。
* **行为定价权**：我们发现，**卖家让步评分（SCS）** 上升与竞价热潮崩塌存在相关性，这标志着卖家议价能力的丧失。

## 关键法医视角研究发现
* **市场堡垒**：基础设施评分（DRS）较高的邮政编码区域能够更快出清过剩库存，验证了“护城河”理论。
* **虚假预警**：该模型识别出一类入市机遇：尽管初始出现供应警报，但10个月后房价反而出现上涨。
* **关键脆弱性**：本次法医审计指出，历史数据中可能存在**生存者偏差（Survivorship Bias）**。