KNN.py

import streamlit as st
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
from sklearn.datasets import make_classification, make_moons, make_circles, make_blobs
from sklearn.model_selection import train_test_split, learning_curve
from sklearn.neighbors import KNeighborsClassifier
from sklearn.metrics import accuracy_score, confusion_matrix, precision_score, recall_score, f1_score, roc_auc_score
from mlxtend.plotting import plot_decision_regions

# image
st.image("https://huggingface.co/spaces/varshitha22/KNN_Algorithm/resolve/main/logo.png")
st.markdown("<br>", unsafe_allow_html=True)

def plot_learning_curves(X_train, y_train, X_test, y_test, model, scoring='accuracy'):
    train_sizes, train_scores, test_scores = learning_curve(model, X_train, y_train, cv=5, scoring=scoring)
    train_mean = np.mean(train_scores, axis=1)
    test_mean = np.mean(test_scores, axis=1)
    
    fig, ax = plt.subplots()
    plt.plot(train_sizes, train_mean, 'o-', color="r", label="Training Score")
    plt.plot(train_sizes, test_mean, 'o-', color="g", label="Cross-validation Score")
    plt.xlabel("Training Examples")
    plt.ylabel("Score")
    plt.legend()
    st.pyplot(fig)

# Sidebar for dataset selection
st.sidebar.header("Dataset Options")
data_type = st.sidebar.selectbox("Select Data Type:", ["Blobs", "Circles", "Moons", "Classification"])
noise = st.sidebar.slider("Add Noise:", 0.0, 1.0, 0.2, step=0.05)

# Sidebar for model selection
st.sidebar.header("Model")
model_name = st.sidebar.radio("Model: ","KNN")

# Display number of neighbors selector only if KNN is selected
if model_name == "KNN":
    neighbors = st.sidebar.number_input("Neighbors", min_value=1, max_value=25, value=5, step=1)
    knn_weights = st.sidebar.radio("KNN Weights:", ["uniform", "distance"])

# KNN Algorithm
st.sidebar.subheader("KNN Algorithm")
algorithms_selected = []
if st.sidebar.checkbox("auto", value=True):
    algorithms_selected.append("auto")
if st.sidebar.checkbox("ball_tree"):
    algorithms_selected.append("ball_tree")
if st.sidebar.checkbox("kd_tree"):
    algorithms_selected.append("kd_tree")
if st.sidebar.checkbox("brute"):
    algorithms_selected.append("brute")

# KNN Metric
st.sidebar.subheader("KNN Metric")
metrics_selected = []
if st.sidebar.checkbox("euclidean", value=True):
    metrics_selected.append("euclidean")
if st.sidebar.checkbox("manhattan"):
    metrics_selected.append("manhattan")
if st.sidebar.checkbox("minkowski"):
    metrics_selected.append("minkowski")

# Generate dataset
if data_type == "Blobs":
    X, y = make_blobs(n_samples=5000, centers=2, cluster_std=noise, random_state=27)
elif data_type == "Circles":
    X, y = make_circles(n_samples=5000, noise=noise, factor=0.5, random_state=27)
elif data_type == "Moons":
    X, y = make_moons(n_samples=5000, noise=noise, random_state=27)
else:
    X, y = make_classification(n_samples=5000, n_features=2, n_classes=2, n_informative=2, n_redundant=0, random_state=27)

# Split dataset
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=27)

# Model selection
if model_name == "KNN":
    model = KNeighborsClassifier(n_neighbors=neighbors, weights=knn_weights, algorithm=algorithms_selected[0] if algorithms_selected else 'auto', metric=metrics_selected[0] if metrics_selected else 'minkowski')

# Fit the model
model.fit(X_train, y_train)

# Display performance metrics only for KNN
if model_name == "KNN":
    st.subheader("KNN Model Evaluation Metrics")
    y_pred = model.predict(X_test)

    # Performance metrics calculation and display
    accuracy = accuracy_score(y_test, y_pred)
    st.write(f"Accuracy: {accuracy:.2f}")
    
    precision = precision_score(y_test, y_pred)
    st.write(f"Precision: {precision:.2f}")
    
    recall = recall_score(y_test, y_pred)
    st.write(f"Recall: {recall:.2f}")
    
    f1 = f1_score(y_test, y_pred)
    st.write(f"F1 Score: {f1:.2f}")
    
    auc = roc_auc_score(y_test, model.predict_proba(X_test)[:, 1]) if hasattr(model, "predict_proba") else "N/A"
    st.write(f"AUC Score: {auc:.2f}")

# Plot dataset
st.subheader("Dataset Visualization")
fig, ax = plt.subplots()
sns.scatterplot(x=X[:, 0], y=X[:, 1], hue=y, palette="coolwarm", s=50, edgecolor="k")
st.pyplot(fig)

# Decision Boundary
st.subheader("Decision Boundary")
fig, ax = plt.subplots()
plot_decision_regions(X_train, y_train, clf=model, legend=2)
st.pyplot(fig)

# Learning Curve
st.subheader("Learning Curve")
plot_learning_curves(X_train, y_train, X_test, y_test, model, scoring='accuracy')