KLearn

Introduction

KLearn is a toolbox for machine learning & data mining(developing).

Setup

python setup.py install
pip install numpy,matplotlib

or

pip install klearn

notes: setup with pip may not install the latest version!

Sensor

use

from klearn.sensor import Sensor

sensor = Sensor(x, y, eta, epoch, draw=True)
"""
:param eta: learning rate
:param epoch: maximum number of cycles
:param draw: whether draw a graph in 2D (default False)
"""
[w,b] = sensor.get_paras()

example

x = np.array([(3, 4, 1), (3, 3, 1)]).T
y = np.array([(1, 1, -1)]).T
sensor = Sensor(x, y, 0.3, 20, draw=True)
[w, b] = sensor.get_paras()

KNN

use

from klearn.knn import KNN

knn = KNN(x, tags, test_point, k, draw=True)
"""
:param x: training set
:param tags: tags of training set
:param test_point: point to be classified
:param k: k value in knn
:param draw: whether draw a graph in 2D or 3D(default False)
"""
tag = knn.get_result()

example

n = 1000
x = np.random.randint(-100, 100, size=(3, n))
y = np.random.randint(0, 3, size=(1, n))
test_point = [3, 4, 100]
knn = KNN(x.T, y[0], test_point, 5, draw=True)
tag = knn.get_result()

kdTree

The toolbox also provide kdTree to store data in knn:

use independently

from klearn.kdtree import KDTree

kdtree = KDTree(x)
kdtree.scan_tree()

use in knn

knn = KNN(x, tags, test_point, k, kdtree=True)
"""
:param kdtree: whether use kdtree to store and train your data(default False)
"""

Naive Bayes

use

from klearn.bayes import Bayes

bayes = Bayes(x, tags, test_x, show=True)
"""
:param show: whether show probabilities in training(default False)
"""
tag = bayes.get_result()

exapmle 1

n = 200
dimensions = 5
x = np.random.randint(0, 4, size=(dimensions, n))
y = np.random.randint(0, 3, size=(1, n))
x_test = np.random.randint(0, 4, size=(1, dimensions))
bayes = Bayes(x, y[0], x_test[0])

example 2

x = [[1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 3, 3, 3, 3, 3, 3],
    ['S', 'M', 'M', 'S', 'S', 'S', 'M', 'M', 'L', 'L', 'L', 'M', 'M', 'L', 'L']]
y = [-1, -1, 1, 1, -1, -1, -1, 1, 1, 1, 1, 1, 1, 1, -1]
test_x = [2, 'S']
bayes = Bayes(x, y, test_x, show=True)
print bayes.get_result()

output in console:

prior probabilities: P(Y=1)=0.6 P(Y=-1)=0.4

conditional_probabilities: P(X(0)=1|Y=1)=0.222222222222 P(X(0)=2|Y=1)=0.333333333333 P(X(0)=3|Y=1)=0.444444444444
P(X(1)=S|Y=1)=0.111111111111 P(X(1)=M|Y=1)=0.444444444444 P(X(1)=L|Y=1)=0.444444444444
P(X(0)=1|Y=-1)=0.5 P(X(0)=2|Y=-1)=0.333333333333 P(X(0)=3|Y=-1)=0.166666666667
P(X(1)=S|Y=-1)=0.5 P(X(1)=M|Y=-1)=0.333333333333 P(X(1)=L|Y=-1)=0.166666666667

posterior_probabilities: P(Y=1)P(X(0)=2|Y=1)P(X(1)=S|Y=1)=0.0222222222222 P(Y=-1)P(X(0)=2|Y=-1)P(X(1)=S|Y=-1)=0.0666666666667

-1