Choose your own

Here we choose one algorithm to classify and measure the accuracy. We will initially try k-nearest neighbors.

Let us import necessary stuff first..

import sys
from class_vis import prettyPicture
from prep_terrain_data import makeTerrainData
from sklearn import tree
from sklearn.metrics import accuracy_score
import numpy as np
import pylab as pl

features_train, labels_train, features_test, labels_test = makeTerrainData()

Let us first check if Decision tree works. This gives us a skeleton to work on.

def classify(features_train, labels_train, min_samples=2):

    ### your code goes here--should return a trained decision tree classifer
    X = features_train
    Y = labels_train
    clf = tree.DecisionTreeClassifier(min_samples_split=min_samples)
    clf = clf.fit(X,Y)
    return clf

# train
clf_2 = classify(features_train, labels_train, 2)

# predict
labels_pred_2 = clf_2.predict(features_test)

# accuracy
acc_min_samples_split_2 = accuracy_score(labels_test, labels_pred_2)


print('Accuracy: {}'.format(acc_min_samples_split_2))

Accuracy: 0.912

prettyPicture(clf_2, features_test, labels_test)

Let us next try using k-nearest neighbor from sci-kit

from sklearn.neighbors import KNeighborsClassifier

def classify(features_train, labels_train, n_neighbors=3):

    ### your code goes here--should return a trained decision tree classifer
    X = features_train
    Y = labels_train
    clf = KNeighborsClassifier(n_neighbors)
    clf = clf.fit(X,Y)
    return clf

# train
clf_k = classify(features_train, labels_train, 3)

# predict
labels_pred_k = clf_k.predict(features_test)

# accuracy
acc_k = accuracy_score(labels_test, labels_pred_k)


print('Accuracy: {}'.format(acc_k))

Accuracy: 0.936

prettyPicture(clf_k, features_test, labels_test)

Improve accuracy?

Let us try increasing neighbors..

# train
n_neighbors = 4

clf_k = classify(features_train, labels_train, n_neighbors)

# predict
labels_pred_k = clf_k.predict(features_test)

# accuracy
acc_k = accuracy_score(labels_test, labels_pred_k)


print('Accuracy: {}'.format(acc_k))

Accuracy: 0.94

Beyond 4, accurancy decreases and so is below 3. So this looks optimal for now.

Good. 94% accuracy is good.