import gradio as gr
import matplotlib.pyplot as plt
from sklearn.datasets import make_blobs
from sklearn.cluster import KMeans
from sklearn.preprocessing import StandardScaler

def compute(number_of_samples, number_of_clusters, randomised_state):
    # create synthetic data
    if randomised_state: seed = None 
    else: seed = 42
    features, true_labels = make_blobs(n_samples=number_of_samples, centers=number_of_clusters, cluster_std=2.75, random_state=seed)
    scaler = StandardScaler()
    scaled_features = scaler.fit_transform(features)
    kmeans = KMeans(init="random", n_clusters=number_of_clusters, n_init=10, max_iter=300, random_state=seed)
    kmeans.fit(scaled_features)

    fig = plt.figure()
    ax1 = fig.add_subplot(111)
    ax1.scatter(scaled_features[:,0], scaled_features[:,1], c=kmeans.labels_, s=8, marker="o", label='dataset')
    ax1.scatter(kmeans.cluster_centers_[:,0],kmeans.cluster_centers_[:,1], s=50, c='r', marker="o", label='cluster centers')
    plt.legend(loc='upper left')
    return fig

iface = gr.Interface(fn=compute, 
                     inputs=[gr.Number(minimum=100, maximum=900, value=100, label="Number of Samples", info="Number of samples must be between 100 and 900"), 
                             gr.Number(minimum=2, maximum=20, value=2, label="Number of Clusters", info="Number of clusters must range between 2 and 20"), 
                             gr.Checkbox(label="randomise samples")], 
                        outputs=gr.Plot(label="Plot"))
iface.launch()