import streamlit as st

st.set_page_config(page_title="KNN", page_icon="🤖", layout="wide")

# Styling - Removed background color, kept font styles
st.markdown("""
    <style>
        h1, h2, h3 {
            color: #003366;
        }
        .custom-font, p {
            font-family: 'Arial', sans-serif;
            font-size: 18px;
            line-height: 1.6;
        }
    </style>
    """, unsafe_allow_html=True)

# Title
st.markdown("<h1 style='color: #003366;'>Understanding K-Nearest Neighbors (KNN)</h1>", unsafe_allow_html=True)
st.image("https://cdn-uploads.huggingface.co/production/uploads/66be28cc7e8987822d129400/7vQBEhJAO_Bju6_SsK0Ms.png")

# Introduction
st.write("""
K-Nearest Neighbors (KNN) is a basic yet powerful machine learning algorithm used for both **classification** and **regression** tasks. It makes predictions by looking at the 'K' closest data points in the training set.

### Key Characteristics:
- KNN is a **non-parametric** and **instance-based** algorithm.
- It **stores** the training data instead of learning a function from it.
- Predictions are made based on **similarity** (distance metrics like Euclidean).
""")

# Working of KNN
st.markdown("<h2 style='color: #003366;'>How KNN Works</h2>", unsafe_allow_html=True)

st.subheader("Training Phase")
st.write("""
- There is **no actual training** involved.
- The algorithm simply memorizes the training data.
""")

st.subheader("Prediction Phase (Classification)")
st.write("""
1. Select a value for **K** (number of neighbors).
2. Measure distances between the new point and training samples.
3. Identify the **K nearest points**.
4. Assign the class based on **majority vote**.
""")

st.subheader("Prediction Phase (Regression)")
st.write("""
1. Choose a value for **K**.
2. Compute distances to training data.
3. Select the **K closest neighbors**.
4. Predict the output as the **average** (or weighted average) of these neighbors' values.
""")

# Overfitting vs Underfitting
st.markdown("<h2 style='color: #003366;'>Model Fit: Overfitting vs Underfitting</h2>", unsafe_allow_html=True)
st.write("""
- **Overfitting**: Happens when K is too low (e.g., K=1); the model becomes too sensitive to noise.
- **Underfitting**: Happens when K is too high; model may miss important patterns.
- **Optimal Fit**: Requires selecting a K value that provides a good balance between bias and variance.
""")

# Training vs CV Error
st.markdown("<h2 style='color: #003366;'>Training vs Cross-Validation Error</h2>", unsafe_allow_html=True)
st.write("""
To choose the best `K`, monitor both:
- **Training Error**: Error on the training set.
- **Cross-Validation (CV) Error**: Error on a validation set, helps assess generalization.

High training accuracy but poor CV accuracy = overfitting.  
Low training and CV accuracy = underfitting.
""")

# Hyperparameter tuning
st.markdown("<h2 style='color: #003366;'>KNN Hyperparameters</h2>", unsafe_allow_html=True)
st.write("""
Main parameters to tune:
- `n_neighbors` (K value)
- `weights`: 'uniform' or 'distance'
- `metric`: Distance metric, e.g., 'euclidean', 'manhattan'
- `n_jobs`: Use multiple processors for speed

These can be optimized using Grid Search, Random Search, or Bayesian methods.
""")

# Feature Scaling
st.markdown("<h2 style='color: #003366;'>Why Feature Scaling is Crucial</h2>", unsafe_allow_html=True)
st.write("""
KNN relies on distance between points, so features must be on the same scale.  
Options:
- **Normalization** (MinMax Scaling): Range [0, 1]
- **Standardization** (Z-score): Mean 0, Std 1

**Important**: Apply scaling *after* splitting your data to avoid data leakage.
""")

# Weighted KNN
st.markdown("<h2 style='color: #003366;'>Weighted KNN</h2>", unsafe_allow_html=True)
st.write("""
In Weighted KNN, closer neighbors contribute more to the prediction.  
This is especially useful when nearby data points are more reliable than distant ones.
""")

# Decision regions
st.markdown("<h2 style='color: #003366;'>Decision Boundaries</h2>", unsafe_allow_html=True)
st.write("""
- K=1 produces sharp, complex boundaries → risk of overfitting.
- Larger K smoothens the boundary → reduces variance but increases bias.
""")

# Cross-validation
st.markdown("<h2 style='color: #003366;'>Understanding Cross-Validation</h2>", unsafe_allow_html=True)
st.write("""
Cross-validation helps evaluate how well the model generalizes.  
**K-Fold Cross Validation**:
- Split data into K parts.
- Train on K-1 parts, test on the remaining.
- Repeat K times and average the performance.
""")

# Hyperparameter search methods
st.markdown("<h2 style='color: #003366;'>Hyperparameter Tuning Methods</h2>", unsafe_allow_html=True)
st.write("""
- **Grid Search**: Tests all combinations — reliable but slow.
- **Random Search**: Randomly samples combinations — faster, may miss optimal.
- **Bayesian Optimization**: Uses past performance to choose next candidates — efficient and smart.
""")

# Link to implementation
st.markdown("<h2 style='color: #003366;'>KNN Code Implementation</h2>", unsafe_allow_html=True)
st.markdown(
    "<a href='https://colab.research.google.com/drive/12lD7ceLj5BPiB6tgxaWXciB1IOMYGyZg#scrollTo=96210031-7967-41c4-9de2-56135c423404' target='_blank' style='font-size: 16px; color: #003366;'>Click here to view the notebook</a>",
    unsafe_allow_html=True
)

# Summary
st.write("""
KNN is a straightforward but effective algorithm.  
To get the best results:
- Scale your data properly.
- Use cross-validation.
- Carefully choose hyperparameters using tuning methods.
""")