varshitha22/KNN_Algorithm

1import streamlit as st2import numpy as np3import matplotlib.pyplot as plt4import seaborn as sns5from sklearn.datasets import make_classification, make_moons, make_circles, make_blobs6from sklearn.model_selection import train_test_split, learning_curve7from sklearn.neighbors import KNeighborsClassifier8from sklearn.metrics import accuracy_score, confusion_matrix, precision_score, recall_score, f1_score, roc_auc_score9from mlxtend.plotting import plot_decision_regions10 11# image12st.image("https://huggingface.co/spaces/varshitha22/KNN_Algorithm/resolve/main/logo.png")13st.markdown("<br>", unsafe_allow_html=True)14 15def plot_learning_curves(X_train, y_train, X_test, y_test, model, scoring='accuracy'):16    train_sizes, train_scores, test_scores = learning_curve(model, X_train, y_train, cv=5, scoring=scoring)17    train_mean = np.mean(train_scores, axis=1)18    test_mean = np.mean(test_scores, axis=1)19    20    fig, ax = plt.subplots()21    plt.plot(train_sizes, train_mean, 'o-', color="r", label="Training Score")22    plt.plot(train_sizes, test_mean, 'o-', color="g", label="Cross-validation Score")23    plt.xlabel("Training Examples")24    plt.ylabel("Score")25    plt.legend()26    st.pyplot(fig)27 28# Sidebar for dataset selection29st.sidebar.header("Dataset Options")30data_type = st.sidebar.selectbox("Select Data Type:", ["Blobs", "Circles", "Moons", "Classification"])31noise = st.sidebar.slider("Add Noise:", 0.0, 1.0, 0.2, step=0.05)32 33# Sidebar for model selection34st.sidebar.header("Model")35model_name = st.sidebar.radio("Model: ","KNN")36 37# Display number of neighbors selector only if KNN is selected38if model_name == "KNN":39    neighbors = st.sidebar.number_input("Neighbors", min_value=1, max_value=25, value=5, step=1)40    knn_weights = st.sidebar.radio("KNN Weights:", ["uniform", "distance"])41 42# KNN Algorithm43st.sidebar.subheader("KNN Algorithm")44algorithms_selected = []45if st.sidebar.checkbox("auto", value=True):46    algorithms_selected.append("auto")47if st.sidebar.checkbox("ball_tree"):48    algorithms_selected.append("ball_tree")49if st.sidebar.checkbox("kd_tree"):50    algorithms_selected.append("kd_tree")51if st.sidebar.checkbox("brute"):52    algorithms_selected.append("brute")53 54# KNN Metric55st.sidebar.subheader("KNN Metric")56metrics_selected = []57if st.sidebar.checkbox("euclidean", value=True):58    metrics_selected.append("euclidean")59if st.sidebar.checkbox("manhattan"):60    metrics_selected.append("manhattan")61if st.sidebar.checkbox("minkowski"):62    metrics_selected.append("minkowski")63 64# Generate dataset65if data_type == "Blobs":66    X, y = make_blobs(n_samples=5000, centers=2, cluster_std=noise, random_state=27)67elif data_type == "Circles":68    X, y = make_circles(n_samples=5000, noise=noise, factor=0.5, random_state=27)69elif data_type == "Moons":70    X, y = make_moons(n_samples=5000, noise=noise, random_state=27)71else:72    X, y = make_classification(n_samples=5000, n_features=2, n_classes=2, n_informative=2, n_redundant=0, random_state=27)73 74# Split dataset75X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=27)76 77# Model selection78if model_name == "KNN":79    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')80 81# Fit the model82model.fit(X_train, y_train)83 84# Display performance metrics only for KNN85if model_name == "KNN":86    st.subheader("KNN Model Evaluation Metrics")87    y_pred = model.predict(X_test)88 89    # Performance metrics calculation and display90    accuracy = accuracy_score(y_test, y_pred)91    st.write(f"Accuracy: {accuracy:.2f}")92    93    precision = precision_score(y_test, y_pred)94    st.write(f"Precision: {precision:.2f}")95    96    recall = recall_score(y_test, y_pred)97    st.write(f"Recall: {recall:.2f}")98    99    f1 = f1_score(y_test, y_pred)100    st.write(f"F1 Score: {f1:.2f}")101    102    auc = roc_auc_score(y_test, model.predict_proba(X_test)[:, 1]) if hasattr(model, "predict_proba") else "N/A"103    st.write(f"AUC Score: {auc:.2f}")104 105# Plot dataset106st.subheader("Dataset Visualization")107fig, ax = plt.subplots()108sns.scatterplot(x=X[:, 0], y=X[:, 1], hue=y, palette="coolwarm", s=50, edgecolor="k")109st.pyplot(fig)110 111# Decision Boundary112st.subheader("Decision Boundary")113fig, ax = plt.subplots()114plot_decision_regions(X_train, y_train, clf=model, legend=2)115st.pyplot(fig)116 117# Learning Curve118st.subheader("Learning Curve")119plot_learning_curves(X_train, y_train, X_test, y_test, model, scoring='accuracy')