KNN을 통한 Parameter Tuning

Assignment 1

KNN으로 HyperParameter 이해하기

우수과제 선정이유

EDA를 통해 IRIS 데이터에 대한 분석과 기본 데이터와 스케일링한 데이터를 비교하며 하이퍼파리머터 튜닝까지 깔끔하게 해 우수과제로 선정되었습니다.

Load Dataset

Import packages

# data
import pandas as pd
import numpy as np
import warnings
warnings.filterwarnings("ignore") 

# visualization
import matplotlib.pyplot as plt
import seaborn as sns
%matplotlib inline
from pandas.plotting import parallel_coordinates

# preprocessing
from sklearn.preprocessing import StandardScaler
from sklearn.preprocessing import MinMaxScaler

# model
from sklearn.neighbors import KNeighborsClassifier
from sklearn.model_selection import train_test_split

# grid search
from sklearn.model_selection import GridSearchCV

# evaluation
from sklearn.model_selection import cross_val_score
from sklearn.metrics import *

Load iris data

from sklearn.datasets import load_iris
# sklearn에 내장되어있는 iris 데이터를 사용
iris = load_iris()
print(iris.DESCR)
# iris dataset 정보를 알 수 있다

  • feature에는 sepal length, sepal width, petal length, petal width가 있다.

  • target은 3개의 class가 있으며 각각 Iris-Setosa, Iris-Versicolour, Iris-Virginica, 즉, 붖꽃의 종류이다.

  • Setosa, Versicolour, Virginica가 각각 0, 1, 2로 분류 되어있다.

  • 총 150개의 instance가 존재한다.

Iris Plants Database
====================

Notes
-----
Data Set Characteristics:
    :Number of Instances: 150 (50 in each of three classes)
    :Number of Attributes: 4 numeric, predictive attributes and the class
    :Attribute Information:
        - sepal length in cm
        - sepal width in cm
        - petal length in cm
        - petal width in cm
        - class:
                - Iris-Setosa
                - Iris-Versicolour
                - Iris-Virginica
    :Summary Statistics:

    ============== ==== ==== ======= ===== ====================
                    Min  Max   Mean    SD   Class Correlation
    ============== ==== ==== ======= ===== ====================
    sepal length:   4.3  7.9   5.84   0.83    0.7826
    sepal width:    2.0  4.4   3.05   0.43   -0.4194
    petal length:   1.0  6.9   3.76   1.76    0.9490  (high!)
    petal width:    0.1  2.5   1.20  0.76     0.9565  (high!)
    ============== ==== ==== ======= ===== ====================

    :Missing Attribute Values: None
    :Class Distribution: 33.3% for each of 3 classes.
    :Creator: R.A. Fisher
    :Donor: Michael Marshall (MARSHALL%PLU@io.arc.nasa.gov)
    :Date: July, 1988

This is a copy of UCI ML iris datasets.
http://archive.ics.uci.edu/ml/datasets/Iris

The famous Iris database, first used by Sir R.A Fisher

This is perhaps the best known database to be found in the
pattern recognition literature.  Fisher's paper is a classic in the field and
is referenced frequently to this day.  (See Duda & Hart, for example.)  The
data set contains 3 classes of 50 instances each, where each class refers to a
type of iris plant.  One class is linearly separable from the other 2; the
latter are NOT linearly separable from each other.

References
----------
   - Fisher,R.A. "The use of multiple measurements in taxonomic problems"
     Annual Eugenics, 7, Part II, 179-188 (1936); also in "Contributions to
     Mathematical Statistics" (John Wiley, NY, 1950).
   - Duda,R.O., & Hart,P.E. (1973) Pattern Classification and Scene Analysis.
     (Q327.D83) John Wiley & Sons.  ISBN 0-471-22361-1.  See page 218.
   - Dasarathy, B.V. (1980) "Nosing Around the Neighborhood: A New System
     Structure and Classification Rule for Recognition in Partially Exposed
     Environments".  IEEE Transactions on Pattern Analysis and Machine
     Intelligence, Vol. PAMI-2, No. 1, 67-71.
   - Gates, G.W. (1972) "The Reduced Nearest Neighbor Rule".  IEEE Transactions
     on Information Theory, May 1972, 431-433.
   - See also: 1988 MLC Proceedings, 54-64.  Cheeseman et al"s AUTOCLASS II
     conceptual clustering system finds 3 classes in the data.
   - Many, many more ...

Make DataFrame

# feature와 target를 하나의 DataFrame으로 만들고 각각의 column명을 붙여주었다.
df = pd.DataFrame(iris.data, columns = iris.feature_names)
y = pd.Series(iris.target, dtype="category")
y = y.cat.rename_categories(iris.target_names)
df['species'] = y
df.head()

sepal length (cm)

sepal width (cm)

petal length (cm)

petal width (cm)

species

0

5.1

3.5

1.4

0.2

setosa

1

4.9

3.0

1.4

0.2

setosa

2

4.7

3.2

1.3

0.2

setosa

3

4.6

3.1

1.5

0.2

setosa

4

5.0

3.6

1.4

0.2

setosa

EDA

df.describe()

sepal length (cm)

sepal width (cm)

petal length (cm)

petal width (cm)

count

150.000000

150.000000

150.000000

150.000000

mean

5.843333

3.054000

3.758667

1.198667

std

0.828066

0.433594

1.764420

0.763161

min

4.300000

2.000000

1.000000

0.100000

25%

5.100000

2.800000

1.600000

0.300000

50%

5.800000

3.000000

4.350000

1.300000

75%

6.400000

3.300000

5.100000

1.800000

max

7.900000

4.400000

6.900000

2.500000

df.groupby('species').size()

  • 각 Class 별로 data가 50개씩 존재한다.

species
setosa        50
versicolor    50
virginica     50
dtype: int64

Pairplot

sns.pairplot(df, hue="species")
plt.show()

  • petal length 만으로 0과 1종을 완전히 구분할 수 있다

  • petal length와 petal width 두가지로 나누면 1과 2종도 구분해 낼 수 있을 것으로 보인다

Distplot

sns.distplot(df[df.species != "setosa"]["petal length (cm)"], hist=True, rug=True, label="setosa")
sns.distplot(df[df.species == "setosa"]["petal length (cm)"], hist=True, rug=True, label="others")
plt.legend()
plt.show()

  • 위의 분포를 보면 petal length 하나의 변수만으로 setosa와 다른 종들을 쉽게 분류해낼 수 있을 것으로 보인다.

sns.distplot(df[df.species == "virginica"]["petal length (cm)"], hist=True, rug=True, label="virginica")
sns.distplot(df[df.species == "versicolor"]["petal length (cm)"], hist=True, rug=True, label="versicolor")
plt.legend()
plt.show()

  • 반면 virginica와 versicolor는 petal length 만으로는 완전히 분류해내기 어려워보인다.

Parallel coordinates plot

parallel_coordinates(df, "species")
plt.xlabel('Features', fontsize=15)
plt.ylabel('Features values', fontsize=15)
plt.legend(loc=1, prop={'size': 15}, frameon=True, shadow=True, facecolor="white", edgecolor="black")
plt.show()

  • 각 feature들로 얼마나 붓꽃 종류를 구분할 수 있는지 한눈에 알 수 있다.

  • 앞에서 언급했듯이 petal length와 petal width로 0과 1를 구분해낼 수 있다

Boxplot¶

df.boxplot()
plt.show()

  • 각 데이터의 분포를 살펴보았다.

  • sepal_width 데이터에서 outlier가 있는 것을 알 수 있었다.

Preprocessing

  • KNN을 사용하기에 앞서 거리를 구하는 것이 분류의 핵심이므로 Scaling을 해주기로 하였다.

  • 그 전에 train data와 test data를 나누어 주기로 하였다.

Split data

# train과 test data를 0.75 : 0.25 비율로 나눈다.
X_train, X_test, y_train, y_test = train_test_split(df.iloc[:,:-1], df.iloc[:,-1], random_state=3)

Scaling

# Standard Scaler
ss = StandardScaler() # Scaling
X_train_s = pd.DataFrame(ss.fit_transform(X_train), columns = X_train.columns)
X_test_s = pd.DataFrame(ss.transform(X_test), columns = X_test.columns)
X_train_s.head()

sepal length (cm)

sepal width (cm)

petal length (cm)

petal width (cm)

0

0.701282

-0.850872

0.852239

0.902670

1

0.444603

-2.033225

0.390008

0.369166

2

0.701282

-0.614401

1.025575

1.169422

3

-0.068753

-0.850872

0.043334

-0.030962

4

-1.608823

1.277364

-1.632254

-1.364723

# Minmax Scaler
ms = MinMaxScaler()
X_train_m = pd.DataFrame(ms.fit_transform(X_train), columns = X_train.columns)
X_test_m = pd.DataFrame(ms.transform(X_test), columns = X_test.columns)
X_train_m.head()

sepal length (cm)

sepal width (cm)

petal length (cm)

petal width (cm)

0

0.583333

0.318182

0.754386

0.750000

1

0.527778

0.090909

0.614035

0.583333

2

0.583333

0.363636

0.807018

0.833333

3

0.416667

0.318182

0.508772

0.458333

4

0.083333

0.727273

0.000000

0.041667

Modeling

Print metrics function

def print_metrics(model, X_train):
    scores = cross_val_score(model, X_train, y_train, cv=10)
    print('*** Cross val score *** \n   {}'.format(scores))
    print('\n*** Mean Accuracy *** \n   {:.7f}'.format(scores.mean()))
    # print('\n*** Confusion Matrix *** \n', confusion_matrix(y_train, model.predict(X_train)))

  • 한번에 validation용 metrics를 출력할 수 있는 함수를 생성하였다.

Simple Model

knn = KNeighborsClassifier()
knn.fit(X_train, y_train)
print_metrics(knn, X_train)
*** Cross val score *** 
   [0.91666667 1.         0.91666667 1.         1.         1.
 1.         1.         1.         0.88888889]

*** Mean Accuracy *** 
   0.9722222

Standard Scaled Model

knn_s = KNeighborsClassifier()
knn_s.fit(X_train_s, y_train)
print_metrics(knn_s, X_train_s)
*** Cross val score *** 
   [0.83333333 1.         0.91666667 1.         0.91666667 1.
 1.         0.90909091 0.8        0.88888889]

*** Mean Accuracy *** 
   0.9264646

MinMax Scaled Model

knn_m = KNeighborsClassifier()
knn_m.fit(X_train_m, y_train)
print_metrics(knn_m, X_train_m)
*** Cross val score *** 
   [0.83333333 1.         0.91666667 1.         0.91666667 1.
 1.         1.         0.9        0.88888889]

*** Mean Accuracy *** 
   0.9455556

Hyperparameter Tuning

Parameters

  • n_neighbors: 검색할 이웃의 수로 default 값은 5이다.

  • Metric: 거리 측정 방식을 변경하는 매개변수로 default 값은 minkowsi이다

  • Weights: 예측에 사용하는 가중치로 uniform 은 각 이웃에 동일한 가중치를 , ‘distance’는 가까운 이웃이 멀리 있는 이웃보다 더욱 큰 영향을 미친다.

Grid Search CV

grid_params = {
    'n_neighbors' : list(range(1,20)),
    'weights' : ["uniform", "distance"],
    'metric' : ['euclidean', 'manhattan', 'minkowski']
}

1. No Scaled

gs = GridSearchCV(knn, grid_params, cv=10)
gs.fit(X_train, y_train)
print("Best Parameters : ", gs.best_params_)
print("Best Score : ", gs.best_score_)
print("Best Test Score : ", gs.score(X_test, y_test))
Best Parameters :  {'metric': 'euclidean', 'n_neighbors': 5, 'weights': 'uniform'}
Best Score :  0.9732142857142857
Best Test Score :  0.9473684210526315

2. Standard Scaled

gs_s = GridSearchCV(knn_s, grid_params, cv=10)
gs_s.fit(X_train_s, y_train)
print("Best Parameters : ", gs_s.best_params_)
print("Best Score : ", gs_s.best_score_)
print("Best Test Score : ", gs_s.score(X_test_s, y_test))
Best Parameters :  {'metric': 'euclidean', 'n_neighbors': 16, 'weights': 'distance'}
Best Score :  0.9732142857142857
Best Test Score :  0.9473684210526315

3. MinMax Scaled

gs_m = GridSearchCV(knn_m, grid_params, cv=10)
gs_m.fit(X_train_m, y_train)
print("Best Parameters : ", gs_m.best_params_)
print("Best Score : ", gs_m.best_score_)
print("Best Test Score : ", gs_m.score(X_test_m, y_test))

  • Score의 결과가 가장 좋은 MinMax Scaled Data의 {'metric': 'euclidean', 'n_neighbors': 9, 'weights': 'uniform'} parameters를 선택하기로 하였다.

Best Parameters :  {'metric': 'euclidean', 'n_neighbors': 9, 'weights': 'uniform'}
Best Score :  0.9732142857142857
Best Test Score :  0.9736842105263158

Final Model

knn_m = KNeighborsClassifier(metric = 'euclidean', n_neighbors = 9, weights = 'uniform')
knn_m.fit(X_train_m, y_train)
print_metrics(knn_m, X_train_m)
*** Cross val score *** 
   [0.91666667 1.         1.         1.         0.91666667 1.
 1.         1.         1.         0.88888889]

*** Mean Accuracy *** 
   0.9722222

Evaluation

def print_test_metrics(model, X_test):
    print('*** Test Accuracy *** \n   {}'.format(model.score(X_test, y_test)))
    print('\n*** Confusion Matrix *** \n', confusion_matrix(y_test, model.predict(X_test)))
print_test_metrics(knn_m, X_test_m)
*** Test Accuracy *** 
   0.9736842105263158

*** Confusion Matrix *** 
 [[15  0  0]
 [ 0 11  1]
 [ 0  0 11]]

  • 하나의 test instance를 제외하고 모두 분류해냈다.

  • 역시 setosa는 완전히 분류해냈으나 나머지 두 종을 분류해내기 힘들었던 것 같다.

Previous데이터 전처리와 시각화NextNavie Bayes 방법론

Last updated

Was this helpful?