기존에 사용하던 변수가 %로 표현되었는데 그 중 간격척도인 온도가 있어서 (원본데이터 - 3시간 이동평균선 데이터)로 바꿔서 측정하였다. 소스코드는 기존과 같고 결과는 아래처럼 나왔다.
========================== Results ==========================
Optimization terminated successfully.
Current function value: 0.389416
Iterations 7
Logit Regression Results
==============================================================================
Dep. Variable: acci No. Observations: 54649
Model: Logit Df Residuals: 54643
Method: MLE Df Model: 5
Date: Tue, 24 Jul 2018 Pseudo R-squ.: -0.1312
Time: 09:23:17 Log-Likelihood: -21281.
converged: True LL-Null: -18812.
LLR p-value: 1.000
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
WindFlow% -0.3271 0.015 -21.267 0.000 -0.357 -0.297
Temp% 0.3170 0.030 10.482 0.000 0.258 0.376
MaxWave% 0.9890 0.096 10.250 0.000 0.800 1.178
SigWave% -1.3945 0.202 -6.897 0.000 -1.791 -0.998
강수량(mm) -0.3900 0.027 -14.631 0.000 -0.442 -0.338
시정(10m) -0.0012 9.29e-06 -133.229 0.000 -0.001 -0.001
==============================================================================
WindFlow% 0.721020
Temp% 1.372957
MaxWave% 2.688644
SigWave% 0.247965
강수량(mm) 0.677048
시정(10m) 0.998764
dtype: float64
West-South
Optimization terminated successfully.
Current function value: 0.528471
Iterations 7
Logit Regression Results
==============================================================================
Dep. Variable: acci No. Observations: 17408
Model: Logit Df Residuals: 17402
Method: MLE Df Model: 5
Date: Tue, 24 Jul 2018 Pseudo R-squ.: 0.001457
Time: 09:23:17 Log-Likelihood: -9199.6
converged: True LL-Null: -9213.0
LLR p-value: 6.109e-05
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
WindFlow% -0.5030 0.026 -19.596 0.000 -0.553 -0.453
Temp% 0.3173 0.066 4.792 0.000 0.188 0.447
MaxWave% 0.2764 0.131 2.113 0.035 0.020 0.533
SigWave% -1.6947 0.258 -6.576 0.000 -2.200 -1.190
강수량(mm) -0.4821 0.058 -8.258 0.000 -0.596 -0.368
시정(10m) -0.0008 1.28e-05 -62.794 0.000 -0.001 -0.001
==============================================================================
WindFlow% 0.604729
Temp% 1.373455
MaxWave% 1.318373
SigWave% 0.183655
강수량(mm) 0.617511
시정(10m) 0.999199
dtype: float64
West-Central
Optimization terminated successfully.
Current function value: 0.366349
Iterations 7
Logit Regression Results
==============================================================================
Dep. Variable: acci No. Observations: 46005
Model: Logit Df Residuals: 45999
Method: MLE Df Model: 5
Date: Tue, 24 Jul 2018 Pseudo R-squ.: -0.1752
Time: 09:23:17 Log-Likelihood: -16854.
converged: True LL-Null: -14341.
LLR p-value: 1.000
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
WindFlow% -0.3007 0.017 -17.711 0.000 -0.334 -0.267
Temp% 0.3072 0.032 9.476 0.000 0.244 0.371
MaxWave% 1.3045 0.113 11.528 0.000 1.083 1.526
SigWave% -1.3484 0.243 -5.553 0.000 -1.824 -0.872
강수량(mm) -0.3935 0.029 -13.405 0.000 -0.451 -0.336
시정(10m) -0.0013 1.06e-05 -123.200 0.000 -0.001 -0.001
==============================================================================
WindFlow% 0.740332
Temp% 1.359617
MaxWave% 3.685696
SigWave% 0.259663
강수량(mm) 0.674694
시정(10m) 0.998696
dtype: float64
South-West
Optimization terminated successfully.
Current function value: 0.455674
Iterations 6
Logit Regression Results
==============================================================================
Dep. Variable: acci No. Observations: 59885
Model: Logit Df Residuals: 59879
Method: MLE Df Model: 5
Date: Tue, 24 Jul 2018 Pseudo R-squ.: 0.03234
Time: 09:23:18 Log-Likelihood: -27288.
converged: True LL-Null: -28200.
LLR p-value: 0.000
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
WindFlow% -0.5744 0.014 -39.922 0.000 -0.603 -0.546
Temp% 0.9198 0.039 23.714 0.000 0.844 0.996
MaxWave% 0.4207 0.070 6.025 0.000 0.284 0.558
SigWave% -0.5422 0.138 -3.921 0.000 -0.813 -0.271
강수량(mm) -0.1299 0.009 -13.877 0.000 -0.148 -0.112
시정(10m) -0.0009 7.12e-06 -132.013 0.000 -0.001 -0.001
==============================================================================
WindFlow% 0.563015
Temp% 2.508860
MaxWave% 1.523075
SigWave% 0.581455
강수량(mm) 0.878226
시정(10m) 0.999061
dtype: float64
South-East
Optimization terminated successfully.
Current function value: 0.483324
Iterations 7
Logit Regression Results
==============================================================================
Dep. Variable: acci No. Observations: 144734
Model: Logit Df Residuals: 144728
Method: MLE Df Model: 5
Date: Tue, 24 Jul 2018 Pseudo R-squ.: 0.007035
Time: 09:23:19 Log-Likelihood: -69953.
converged: True LL-Null: -70449.
LLR p-value: 4.636e-212
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
WindFlow% -0.5170 0.008 -62.105 0.000 -0.533 -0.501
Temp% 0.6344 0.022 28.323 0.000 0.590 0.678
MaxWave% 0.1932 0.034 5.621 0.000 0.126 0.260
SigWave% -0.4455 0.069 -6.475 0.000 -0.580 -0.311
강수량(mm) -0.2044 0.008 -25.115 0.000 -0.220 -0.188
시정(10m) -0.0009 4.29e-06 -198.708 0.000 -0.001 -0.001
==============================================================================
WindFlow% 0.596288
Temp% 1.885815
MaxWave% 1.213066
SigWave% 0.640527
강수량(mm) 0.815153
시정(10m) 0.999149
dtype: float64
East-South
Optimization terminated successfully.
Current function value: 0.505116
Iterations 7
Logit Regression Results
==============================================================================
Dep. Variable: acci No. Observations: 76009
Model: Logit Df Residuals: 76003
Method: MLE Df Model: 5
Date: Tue, 24 Jul 2018 Pseudo R-squ.: -0.005069
Time: 09:23:19 Log-Likelihood: -38393.
converged: True LL-Null: -38200.
LLR p-value: 1.000
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
WindFlow% -0.4887 0.011 -45.692 0.000 -0.510 -0.468
Temp% 0.4496 0.029 15.763 0.000 0.394 0.506
MaxWave% 0.1066 0.042 2.553 0.011 0.025 0.188
SigWave% -0.3673 0.084 -4.366 0.000 -0.532 -0.202
강수량(mm) -0.3801 0.017 -22.214 0.000 -0.414 -0.347
시정(10m) -0.0008 5.62e-06 -138.072 0.000 -0.001 -0.001
==============================================================================
WindFlow% 0.613439
Temp% 1.567751
MaxWave% 1.112502
SigWave% 0.692618
강수량(mm) 0.683776
시정(10m) 0.999225
dtype: float64
East-Central
========================== Results ==========================
서남, 남서쪽 바다의 경우 정답률이 다른 바다보다 낮게 나왔다.
이유는 1. 데이터 정제 과정에서 서남, 남서쪽 데이터가 유실되었고, 2. 다도해가 다른 바다와 다른 특성을 가지기 때문이라고 추측된다.