🏠 NJ Home Price Forecasting using R

📈 This project predicts median housing prices in New Jersey using time series forecasting techniques in R. It applies classical and decomposition-based models to uncover patterns and predict future values in the housing market.

---

📁 Project Overview

• Objective: Forecast median home prices in New Jersey using historical trends
• Tools: R, RMarkdown
• Libraries: fpp, fpp2, TTR, dplyr, ggplot2, forecast

---

🗂️ Table of Contents

1. Data Collection
2. Data Preprocessing
3. Modeling
4. Results
5. Usage
6. Contributing

---

📦 Data Collection

• Source: Median listing prices dataset for all homes in NJ
• Format: .csv file, time-series indexed
• Columns include: Region name, Date, Median Price

---

🔧 Data Preprocessing

• Converted time columns to time-series format
• Handled missing values and removed noise
• Created visual exploratory plots (trends & seasonality)

---

🤖 Modeling Approach

📉 Models Implemented

• Baseline Models:

  • Naive Forecast
  • Moving Average
  • Simple Exponential Smoothing (SES)

• Advanced Model:

  • Holt-Winters Additive
  • Decomposition Additive (Season-Trend)

---

📊 Results Summary

Model	RMSE	MAE	MAPE
Naive	772.625	703.041	700.607
Simple Exp Smoothing	745.526	693.019	772.871
Holt-Winters Additive	611.131	520.891	514.702

✅ Holt-Winters Additive gave the lowest error rates, indicating superior performance in capturing both seasonality and trend.

---

🧠 Key Insights

• 📈 Forecast shows upward housing trend in NJ over the next 2–3 years
• 🧪 Holt-Winters model best captured the trend + seasonality
• 📉 Residual diagnostics confirmed randomness → robust forecasting

---

⚙️ Usage

# Clone the repo
git clone https://github.com/kumarritik24/NJ_Home_Price_Forecasting.git

# Open the .Rmd file in RStudio to run the code and generate output