This README explains the SHAP (SHapley Additive exPlanations) visualizations for an XGBoost model predicting house prices based on the California Housing dataset.
<<<<<<< HEAD
- Simple Neural Network Without Machine Learning Libraries
- Implemented a basic neural network from scratch using only Python's built-in modules.
- SHAP Explanation for XGBoost on California Housing Data
- SHAP visualizations for XGBoost on the California Housing dataset reveal how features like MedInc, Latitude, and Longitude influence house price predictions.
- Gap Statistic Method for Finding Optimal Clusters
- A statistical approach to determine the optimal number of clusters by comparing actual clustering performance with a random baseline.
- Linear & Logistic Regression from Scratch
- Linear Regression predicts continuous values using a straight-line equation optimized by gradient descent, while Logistic Regression performs binary classification by applying the sigmoid function to a linear model.
- Support Vector Machine (SVM) from Scratch
- Implements a Support Vector Machine (SVM) classifier using Stochastic Gradient Descent (SGD)for binary classification. The model is trained using the hinge loss function and L2 regularization.
- Regression Analysis with Scikit-Learn
- Implements various regression techniques using Scikit-Learn and other relevant Python libraries.
- Deep Neural Network from Scratchn
- This project implements two versions of a Deep Neural Network (DNN) from scratch using NumPy =======
The dataset includes features like:
- MedInc (Median Income)
- HouseAge (House Age)
- AveRooms (Average Rooms per House)
- Latitude & Longitude (Location)
- Population & Occupancy
8533f4c (Add readme)
The model predicts house prices using these features.
- How each feature affects a single prediction.
- The blue line traces how features push the price up or down.
- Key Influence: MedInc, Latitude, and Longitude have the biggest impact.
- How a prediction was made for one house.
- Blue pushes the price down, red pushes it up.
- MedInc, Latitude, and AveOccup strongly influence this prediction.
- Overall feature importance across all predictions.
- Red = High feature value, Blue = Low feature value.
- MedInc, Latitude, and Longitude are the most important.
- MedInc (Income) is the strongest predictor of house prices.
- Location (Latitude & Longitude) also plays a big role.
- SHAP makes model decisions explainable and transparent.
🚀 SHAP helps us understand how AI makes predictions!