• This repository contains a data science project focused on understanding the key factors influencing US home prices over the last 20 years.

• The project utilizes the S&P Case-Schiller Home Price Index as a proxy for home prices and explores various economic indicators to build a predictive model.

• The chosen model is a Random Forest Regressor, achieving a remarkable R2 score of 99.87%

• The objective is to identify and analyze factors that significantly impact home prices in the United States.

• By leveraging publicly available data, we aim to build a robust predictive model that explains the variations in the S&P Case-Schiller Home Price Index over the past two decades.

The following features were selected for the predictive model:

• Median market value of all homes.
• Reflects overall trends in home values, providing insights into market conditions and price movements.

• Reflects the average cost of newly constructed homes.
• Influences perceptions of the affordability of new housing.

• Provides insights into the typical price of newly sold houses.
• Helps understand the distribution of new home prices.

• Reflects the level of investment in residential construction.
• Influences housing supply, potentially affecting home prices.

• Influences economic conditions and affects consumer confidence in buying houses.

• Indicates inflation in housing costs, potentially impacting home prices.

• Population growth influences housing demand, potentially affecting home prices.

• Property taxes influence the overall cost of living.

• Authorized housing units can influence home prices and increase demand.

• Reflects the balance between housing supply and demand.

• Provides insights about a fixed interest rate for 30 years that buyers can afford for a loan.

• Provides insights into consumer sentiment about the economy and the housing market.

• Influences job security; more job security improves purchasing ability.

• Provides insights into vacant unit availability, market conditions, and supply and demand balance.

• Changing in the federal reserve; adjusts the federal funds rate may influence the mortgage rate.

• Utilized machine learning models to impute missing values in the dataset.

• Conducted thorough EDA to understand the relationships between features and the target variable.

• Engineered relevant features to improve model performance.

• Chose the Random Forest Regressor based on its outstanding R2 score of 99.87%.

• Utilized R2 score, Mean Absolute Error (MAE), and Root Mean Squared Error (RMSE) for model evaluation.

• After training the Random Forest Regressor model, the feature importance analysis has done to reveal the predictor variables that contributes more when predicting target variable.

• Programming Langauge : Python

• Code Notebook : Google Colab-notebook

• Data Collection: FRED (Federal Reserve Economic Data)

• Data Cleaning: Machine Learning Imputation

• Exploratory Data Analysis: Pandas, Matplotlib, Seaborn

• Model Building: Scikit-Learn (Random Forest Regressor)

• Model Evaluation: R2 Score, MAE, RMSE

• Model Fit (R2 Score): 99.87%
• MAE (Mean Absolute Error): 1.39
• RMSE (Root Mean Squared Error): 2.37
• The Random Forest Regressor demonstrated superior performance in minimizing errors and capturing the variance in the target variable compared to other models