Iris-Dataset_Subplots

# This program aims to find which combination of features from the iris dataset will create the most distinct groups to use for KNN

from sklearn.datasets import load_iris
import numpy as np
import matplotlib.pyplot as plt
# load data, cast as numpy array
iris = load_iris()
arr = np.array(iris.data)
# separate the data into the three types of irises
setosa = arr[:50]
versicolor = arr[50:100]
virginica = arr[100:]
# function: creates coordinates based on the features given
target = {"sepal_length": 0, "sepal_width": 1, "petal_length": 2, "petal_width": 3}
def points(flower, first_char, second_char):
    x = []
    y = []
    for i in flower:
        x.append(i[target[first_char]])
        y.append(i[target[second_char]])
    return [x, y]
# experiment with which combination of features makes the most distinct groups on the scatterplot (6 possible combinations)
combos = [["sepal_length", "sepal_width"],
          ["sepal_length", "petal_length"],
          ["sepal_length", "petal_width"],
          ["sepal_width", "petal_length"],
          ["sepal_width", "petal_width"],
          ["petal_length", "petal_width"]]
# create subplots to fit all of the six scatterplots together
name_arrays = {"setosa": setosa, "versicolor": versicolor, "virginica": virginica}
fig, axs = plt.subplots(2, 3)
row = 0
column = 0
for combo in combos:
    for string, array in name_arrays.items():
        coord = points(array, combo[0], combo[1])
        graph = axs[row-1][column-1]
        graph.scatter(coord[0], coord[1], label=string if combos.index(combo) == 0 else "")
        graph.set_title(combo[0] + " by " + combo[1])
        graph.legend()
    column += 1
    if column == 3:
        row = 1
        column = 0
plt.suptitle("Comparison of Iris Features")
plt.show()
plt.savefig("iris_classification_char_comparison.png")