003_제3유형_구름 기반 실습

import pandas as pd
from scipy.stats import chi2_contingency

# Load the Titanic dataset
df = pd.read_csv("data/Titanic.csv")

# Create a contingency table for 'Gender' and 'Survived'
table = pd.crosstab(df['Gender'], df['Survived'])

# Perform the chi-squared test
statistics, p, degrees_of_freedom, expected = chi2_contingency(table)

# Print the chi-squared statistic rounded to 3 decimal placesprint(round(statistics, 3))

각 단계 설명:
데이터 로드: pd.read_csv("data/Titanic.csv")는 Titanic 데이터 세트를 DataFrame으로 읽습니다.

비상사태 테이블 생성: pd.crosstab(df['Gender'], df['Survived'])는 각 성별에 대한 생존자와 비생존자의 수를 보여주는 테이블을 생성합니다.

 

카이제곱 테스트: chi2_contingency(table)는 분할표에 대해 카이제곱 테스트를 수행합니다. 이 함수는 다음을 반환합니다.
통계: 카이제곱 검정 통계량,

 

p: p-값,

 

자유도: 자유도,

 

예상: 변수 사이에 연관성이 없는 경우 예상 빈도표입니다.
인쇄: print(round(statistics, 3))는 테스트 통계를 소수점 이하 3자리로 반올림하여 출력

from statsmodels.api import Logit
df['Gender'] = df['Gender'].map({'male': 0, 'female': 1})
x = df[['Gender','SibSp','Parch','Fare']]
y = df['Survived']
model =Logit(y, x)
result = model.fit()
print(result.summary())

Optimization terminated successfully.          Current function value: 0.610661          Iterations 6                            Logit Regression Results                            ============================================================================== Dep. Variable:               Survived   No. Observations:                  891 Model:                          Logit   Df Residuals:                      887 Method:                           MLE   Df Model:                            3 Date:                Sun, 10 Nov 2024   Pseudo R-squ.:                 0.08297 Time:                        14:33:06   Log-Likelihood:                -544.10 converged:                       True   LL-Null:                       -593.33 Covariance Type:            nonrobust   LLR p-value:                 3.335e-21 ==============================================================================                  coef    std err          z      P>|z|      [0.025      0.975] ------------------------------------------------------------------------------ Gender         1.6051      0.169      9.522      0.000       1.275       1.935 SibSp         -0.5659      0.102     -5.562      0.000      -0.765      -0.367 Parch         -0.2331      0.105     -2.224      0.026      -0.438      -0.028 Fare           0.0013      0.002      0.892      0.373      -0.002       0.004 ============================================================================== 프로세스가 종료되었습니다.

 

v1.0

v2.0

import statsmodels.api as sm df['Gender'] = df['Gender'].map({'male': 0, 'female': 1}) # Select the predictor variables x = df[['Gender', 'SibSp', 'Parch', 'Fare']] # Select the target variable y = df['Survived'] # Add a constant to the predictor variables (for the intercept term in logistic regression) x = sm.add_constant(x) # Fit the logistic regression model model = sm.Logit(y, x) result = model.fit() # Print the summary of the model print(result.summary())

import statsmodels.api as sm from sklearn.preprocessing import LabelEncoder encoder = LabelEncoder() df['Gender'] = encoder.fit_transform(df['Gender']) x = df[['Gender','SibSp','Parch','Fare']] y = df['Survived'] x = sm.add_constant(x) model = sm.Logit(y, x) result = model.fit() print(result.summary())

 

 

#문제 2번

이 코드는 'statsmodels'를 사용하여 로지스틱 회귀를 수행하여 특정 변수가 생존 확률에 미치는 영향을 분석합니다

(Titanic 데이터 세트 변수 사용). 각 부분에 대한 분석은 다음과 같습니다.
import statsmodels.api as sm
from sklearn.preprocessing import LabelEncoder

# Encode 'Gender' as numeric
encoder = LabelEncoder()
df['Gender'] = encoder.fit_transform(df['Gender'])
라이브러리 가져오기: 'statsmodels'는 로지스틱 회귀 모델에 사용되고 'sklearn'의 'LabelEncoder'는 범주형 변수를 숫자 형식으로 변환하는 데 사용됩니다.
인코딩 성별: LabelEncoder는 범주형 성별 열(예: "남성", "여성")을 숫자 값(예: 0 및 1)으로 변환합니다. 로지스틱 회귀에는 수치 입력이 필요하므로 이는 필수적입니다.
x = df[['Gender', 'SibSp', 'Parch', 'Fare']]
y = df['Survived']
예측 변수(x) 선택: 코드는 열이 있는 DataFrame인 x를 생성합니다.

Gender: 0 또는 1로 인코딩됩니다.
SibSp: 탑승한 형제자매/배우자의 수.
Parch: 탑승한 부모/자녀 수.
운임: 여객운임.
타겟 변수(y) 선택: 타겟 변수 y를 Survived로 설정하여 승객이 생존했는지(1), 그렇지 않았는지(0)를 나타냅니다.

x = sm.add_constant(x)
상수 추가(절편): sm.add_constant(x)는 x에 절편 항을 추가하여 로지스틱 회귀가 모델에 상수를 포함할 수 있도록 합니다.
model = sm.Logit(y, x)
result = model.fit()
로지스틱 회귀 모델 생성 및 적합:
model = sm.Logit(y, x)는 y를 종속 변수로, x를 독립 변수로 사용하여 로지스틱 회귀 모델을 초기화합니다.
result = model.fit()은 모델을 데이터에 맞춰 각 예측 변수에 대한 계수를 추정합니다.
print(result.summary())
모델 요약 인쇄: result.summary()는 다음을 포함하는 자세한 요약을 제공합니다.
생존 확률에 미치는 영향을 나타내는 각 예측 변수의 계수입니다.
가설 검정을 위한 p-값.
모델 성능을 평가하기 위한 로그 우도, 유사 R-제곱 등과 같은 적합도 측정항목.
이 로지스틱 회귀 모델을 사용하면 각 변수(Gender, SibSp, Parch, Fare)가 타이타닉호의 생존 확률에 어떻게 영향을 미치는지 이해할 수 있습니다.

 

Optimization terminated successfully.
         Current function value: 0.482065
         Iterations 6
                           Logit Regression Results                           
==============================================================================
Dep. Variable:               Survived   No. Observations:                  891
Model:                          Logit   Df Residuals:                      886
Method:                           MLE   Df Model:                            4
Date:                Sun, 10 Nov 2024   Pseudo R-squ.:                  0.2761
Time:                        14:39:08   Log-Likelihood:                -429.52
converged:                       True   LL-Null:                       -593.33
Covariance Type:            nonrobust   LLR p-value:                 1.192e-69
==============================================================================
                 coef    std err          z      P>|z|      [0.025      0.975]
------------------------------------------------------------------------------
const          0.9466      0.169      5.590      0.000       0.615       1.279
Gender        -2.6422      0.186    -14.197      0.000      -3.007      -2.277
SibSp         -0.3539      0.098     -3.604      0.000      -0.546      -0.161
Parch         -0.2007      0.112     -1.792      0.073      -0.420       0.019
Fare           0.0147      0.003      5.553      0.000       0.010       0.020
==============================================================================

import statsmodels.api as sm
from sklearn.preprocessing import LabelEncoder
encoder = LabelEncoder()
df['Gender'] = encoder.fit_transform(df['Gender'])
formula = "Survived ~ Gender + SibSp + Parch + Fare"
result = sm.Logit.from_formula(formula, df).fit()
print(result.summary())

반응형

#문제 3번

import numpy as np

print(round(np.exp(result.params['SibSp']),3))
print(round(np.exp(-0.3539),3))
0.702
0.702