Recap: City Lights

Part of the Unsupervised Learning section of Coddy's Machine Learning journey. Lesson 23 of 51.

Challenge

Medium

Remember the valley map from DBSCAN? Now you make one. A satellite photographed a valley at night; group its lights into towns and find the lone ones: farms, a lighthouse, a camper. The program reads eps in km.

Run DBSCAN with that eps and min_samples=5.
Print the number of towns, the number of lone lights, and the biggest town's size.
Map them: each town its own scatter labelled town of N lights, lone lights as gray x marks labelled lone lights, a legend. Show it.

For another valley and eps 1.3:

Towns: 3
Lone lights: 10
Biggest town: 77 lights

A chart titled Valley at Night: a long curved town of 77 lights, a round town of 47 and a village of 20, each in its own color, and 10 lone lights as gray x marks, with a legend

Coddy describes your map. Then try eps 5: which towns merge?

Try it yourself

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from sklearn.cluster import DBSCAN

# A satellite's night photo of a valley: one row per light, position in km
lights = pd.read_csv("lights.csv")
X = np.column_stack([lights["east_km"], lights["north_km"]])
eps = float(input())

fig, ax = plt.subplots()
ax.set_title("Valley at Night")
ax.set_xlabel("East (km)")
ax.set_ylabel("North (km)")

# Write your code below