slope_cluster.py

import numpy as np
import random
import matplotlib.pyplot as plt


def clustering(x1, x2):
    # randomly initialize centroids
    c1 = [random.randint(10, 40), random.randint(10, 40)]
    c2 = [random.randint(50, 100), random.randint(50, 100)]
    m = len(x1)
    labels = [""] * m

    for t in range(20000):
        prev_labels = labels
        for i in range(m):
            x = [x1[i], x2[i]]

            # calculate the distance between the current x and centroids
            distance_to_c1 = np.sqrt(np.square(c1[0] - x[0]) + np.square(c1[1] - x[1]))
            distance_to_c2 = np.sqrt(np.square(c2[0] - x[0]) + np.square(c2[1] - x[1]))

            if distance_to_c1 > distance_to_c2:
                labels[i] = "dangerous"
            else:
                labels[i] = "safe"

        # move the centroids
        sum1 = [0, 0]
        sum2 = [0, 0]
        for i in range(m):
            x = [x1[i], x2[i]]
            if labels[i] == 0:
                sum1[0] += x[0]
                sum1[1] += x[1]
            else:
                sum2[0] += x[0]
                sum2[1] += x[1]
        c1[0] = sum1[0] / len(labels)
        c1[1] = sum1[1] / len(labels)
        c2[0] = sum2[0] / len(labels)
        c2[1] = sum2[1] / len(labels)

        # check if labels converged
        changed = False
        for i in range(len(labels)):
            if labels[i] != prev_labels[i]:
                changed = True
        if not changed:
            break
    return labels


# slope degrees: for ex. 27°
slopes = [77, 33, 75, 9, 72, 34, 73, 21, 99, 14,
          92, 20, 55, 19, 75, 30, 70, 0, 93, 37,
          98, 39, 80, 32, 78, 39, 89, 37, 55, 8,
          65, 15, 63, 7, 57, 10, 82, 12, 70, 39]

# average precipitation: for ex. 60%
precis = [81, 32, 56, 45, 91, 37, 93, 22, 82, 3,
          70, 26, 57, 1, 59, 22, 67, 25, 97, 30,
          70, 8, 58, 13, 91, 34, 81, 34, 83, 28,
          91, 3, 82, 35, 72, 35, 92, 3, 87, 6]

labels = clustering(slopes, precis)
safe_slopes = [[], []]
dangerous_slopes = [[], []]

for i in range(len(labels)):
    if labels[i] == "safe":
        safe_slopes[0].append(slopes[i])
        safe_slopes[1].append(precis[i])
    else:
        dangerous_slopes[0].append(slopes[i])
        dangerous_slopes[1].append(precis[i])

scatter_c1 = plt.scatter(safe_slopes[0], safe_slopes[1], c='green', marker='o')
scatter_c2 = plt.scatter(dangerous_slopes[0], dangerous_slopes[1], c='red', marker='x')
plt.legend((scatter_c1, scatter_c2), ("Safe", "Dangerous"),
           scatterpoints=1,
           loc='lower right', ncol=3,
           fontsize=10)

plt.title("Clustering cut-slopes")
plt.ylabel("Slope degrees")
plt.xlabel("Average precipitation (yearly)")
plt.grid()
plt.show()