🧠 Artificial Neuron (Perceptron) - Rent Estimator

A from-scratch implementation of an Artificial Neuron (Perceptron) in C# without using any external Machine Learning libraries. This project models the relationship between house features (Number of Rooms, Distance to Center, Building Age) and their Rent prices using supervised learning.

🎯 Project Overview

This project demonstrates the fundamental mechanics of a neural network: input processing, weight initialization, activation functions, and backpropagation (weight updates). The goal is to predict house rents based on a small normalized dataset.

📊 The Dataset (Normalized Summary)

The model is trained on 15 samples and tested on 5 samples. Before processing, the data is normalized (e.g., Rooms/5, Distance/20, Age/30, Rent/10000) to keep values between 0 and 1.

Feature	Description	Normalization Factor
Input 1	Number of Rooms	/ 5
Input 2	Distance to Center (km)	/ 20
Input 3	Building Age (years)	/ 30
Target	Rent (₺)	/ 10,000

🧮 Mathematical Model

This implementation explicitly programs the core math of a single neuron:

1. Summation Function ($v$): Calculates the weighted sum: $v = \sum_{i=1}^{n} w_i x_i$
2. Activation Function (Sigmoid): Maps the output between 0 and 1: $y = \frac{1}{1 + e^{-v}}$
3. Weight Update Rule: Adjusts weights based on error ($t - y$) and learning rate ($\lambda$): $w_i = w_i + \lambda(t - y)x_i$
4. Error Evaluation (MSE): Calculates the Mean Squared Error: $MSE = \frac{1}{n} \sum (t - y)^2$

🔬 Experimental Analysis

The program automatically runs multiple simulations to analyze how different Hyperparameters affect the model's performance:

• Feature Selection: Compares 3-input vs. 2-input (excluding Building Age) performance.
• Learning Rates ($\lambda$): Tests $0.05$, $0.01$, and $0.1$ to find the best convergence.
• Epochs: Compares training results after $25$ and $100$ cycles.

A final summary table is printed to the console, allowing for a detailed comparison of Mean Squared Error (MSE) across all test cases.

🛠️ How to Run

1. Clone the repository.
2. Open the .sln file in Visual Studio.
3. Build and Run to see step-by-step training progress and final evaluation tables.