Skip to content

README.md

Latest commit

 

History

History
59 lines (46 loc) · 2.94 KB

README.md

File metadata and controls

59 lines (46 loc) · 2.94 KB

House Price Analysis

This directory contains an exploratory data analysis (EDA) of the Ames Housing dataset, focusing on house price insights through statistical summaries and visualizations. No machine learning models for prediction are implemented; the focus is on data understanding, feature correlations, and outlier identification.

Files

  • House.ipynb: Jupyter Notebook with the complete analysis.
  • House_prediction.csv: Dataset file (Ames Housing data with features like lot area, living area, quality ratings, and sale prices).

Dataset Overview

  • Shape: 1460 rows, 81 columns.
  • Target: SalePrice (house sale price in USD).
  • Key features analyzed: OverallQual, GrLivArea, GarageCars, GarageArea, TotalBsmtSF, LotArea, neighborhoods, etc.

Methods and Techniques Used

  1. Data Loading and Preparation:

    • Load data using pandas.read_csv().
    • Suppress warnings and set plotting styles with matplotlib and seaborn.

  2. Dataset Inspection:

    • Display shape, memory usage, first rows (df.head()).
    • Column info (df.info()), descriptive stats (df.describe()).
    • Identify numerical and categorical columns.

  3. Missing Values Analysis:

    • Calculate missing counts and percentages.
    • Display as a sorted DataFrame for columns with missing data (e.g., PoolQC at 99.5% missing).

  4. Statistical Analysis:

    • Correlation matrix for numerical features using df.corr().
    • Focus on correlations with SalePrice (e.g., OverallQual: 0.791).
    • Skewness and kurtosis using scipy.stats.skew() and kurtosis().
    • Grouped statistics by neighborhood and overall quality using groupby() and aggregations (mean, median, etc.).

  5. Visualizations:

    • Histogram and KDE for SalePrice distribution (seaborn.histplot()).
    • Boxplots for SalePrice by neighborhood and quality (seaborn.boxplot()).
    • Scatterplots for key correlations (e.g., GrLivArea vs. SalePrice) using seaborn.scatterplot().
    • Correlation heatmap (seaborn.heatmap()).
    • Pairplot for selected features (seaborn.pairplot()).

  6. Outlier Detection:

    • Use IQR method: Calculate Q1, Q3, and bounds (1.5 * IQR).
    • Applied to SalePrice, GrLivArea, LotArea, TotalBsmtSF.
    • Results summarized in a DataFrame (e.g., 4.2% outliers in SalePrice).

  7. Insights Summary:

    • Printed key findings: averages, medians, top correlations, neighborhood insights, quality breakdowns, data quality notes, and modeling recommendations (e.g., log transform SalePrice, handle missing values).

Key Insights

  • Average sale price: $180,921.
  • Strongest predictors: Overall quality and living area.
  • Outliers: 61 in sale price (4.2%).
  • Recommendations: Focus on key features for future modeling, engineer new ones like house age.

Usage

Open House.ipynb in Jupyter and run cells sequentially. Ensure House_prediction.csv is in the same directory.

Dependencies

  • pandas, numpy, matplotlib, seaborn, scipy.