[ECC DS 11주차] 회귀 3_Pycaret Introduction

0. Pycaret

0-1. AutoML

• AutoML(Automated Machine Learning)

  • 현재의 머신러닝 모델링은 Machine Learning Process 동안 많은 시간과 노력이 요구됨

  • AutoML은 기계 학습 파이프라인에서 수작업과 반복되는 작업을 자동화하는 프로세스

  • 머신러닝을 자동화하는 AI 기술

cf) Machine Learning Process: 문제 정의 과정, 데이터 수집, 전처리, 모델 학습 및 평가, 서비스 적용

cf) 파이프라인: 한 데이터 처리 단계의 출력이 다음 단계의 입력으로 이어지는 형태로 연결된 구조

0-2. Pycaret

• Low-code machine learning

• AutoML을 가능하게 해주는 파이썬 라이브러리

• scikit-learn 패키지 기반

• 분류, 회귀, 군집화 등 다양한 모델 지원

cf) low-code: 어플리케이션과 시스템을 빌딩할 때 거의 코딩이 필요 없는 방식의 소프트웨어

---

- setup

- compare models

- create and store models in variable

- blend models

- stack models


# **1. 데이터 준비**

- 데이터는 이전 예제에서 다룬 house price 데이터를 활용

! pip install pycaret

```python
### Import libraries

import numpy as np 
import pandas as pd 

from pycaret.regression import * # pycaret에서 회귀 관련 모듈 import

### Import dataset

train = pd.read_csv('/content/drive/MyDrive/Colab Notebooks/ECC 48기 데과B/10주차/data/house_price_train.csv')
test  = pd.read_csv('/content/drive/MyDrive/Colab Notebooks/ECC 48기 데과B/10주차/data/house_price_test.csv')
sample= pd.read_csv('/content/drive/MyDrive/Colab Notebooks/ECC 48기 데과B/10주차/data/house_sample_submission.csv')

train.head()

	Id	MSSubClass	MSZoning	LotFrontage	LotArea	Street	Alley	LotShape	LandContour	Utilities	...	PoolArea	PoolQC	Fence	MiscFeature	MiscVal	MoSold	YrSold	SaleType	SaleCondition	SalePrice
0	1	60	RL	65.0	8450	Pave	NaN	Reg	Lvl	AllPub	...	0	NaN	NaN	NaN	0	2	2008	WD	Normal	208500
1	2	20	RL	80.0	9600	Pave	NaN	Reg	Lvl	AllPub	...	0	NaN	NaN	NaN	0	5	2007	WD	Normal	181500
2	3	60	RL	68.0	11250	Pave	NaN	IR1	Lvl	AllPub	...	0	NaN	NaN	NaN	0	9	2008	WD	Normal	223500
3	4	70	RL	60.0	9550	Pave	NaN	IR1	Lvl	AllPub	...	0	NaN	NaN	NaN	0	2	2006	WD	Abnorml	140000
4	5	60	RL	84.0	14260	Pave	NaN	IR1	Lvl	AllPub	...	0	NaN	NaN	NaN	0	12	2008	WD	Normal	250000

5 rows × 81 columns

### 데이터 정보 확인

train.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 1460 entries, 0 to 1459
Data columns (total 81 columns):
 #   Column         Non-Null Count  Dtype  
---  ------         --------------  -----  
 0   Id             1460 non-null   int64  
 1   MSSubClass     1460 non-null   int64  
 2   MSZoning       1460 non-null   object 
 3   LotFrontage    1201 non-null   float64
 4   LotArea        1460 non-null   int64  
 5   Street         1460 non-null   object 
 6   Alley          91 non-null     object 
 7   LotShape       1460 non-null   object 
 8   LandContour    1460 non-null   object 
 9   Utilities      1460 non-null   object 
 10  LotConfig      1460 non-null   object 
 11  LandSlope      1460 non-null   object 
 12  Neighborhood   1460 non-null   object 
 13  Condition1     1460 non-null   object 
 14  Condition2     1460 non-null   object 
 15  BldgType       1460 non-null   object 
 16  HouseStyle     1460 non-null   object 
 17  OverallQual    1460 non-null   int64  
 18  OverallCond    1460 non-null   int64  
 19  YearBuilt      1460 non-null   int64  
 20  YearRemodAdd   1460 non-null   int64  
 21  RoofStyle      1460 non-null   object 
 22  RoofMatl       1460 non-null   object 
 23  Exterior1st    1460 non-null   object 
 24  Exterior2nd    1460 non-null   object 
 25  MasVnrType     1452 non-null   object 
 26  MasVnrArea     1452 non-null   float64
 27  ExterQual      1460 non-null   object 
 28  ExterCond      1460 non-null   object 
 29  Foundation     1460 non-null   object 
 30  BsmtQual       1423 non-null   object 
 31  BsmtCond       1423 non-null   object 
 32  BsmtExposure   1422 non-null   object 
 33  BsmtFinType1   1423 non-null   object 
 34  BsmtFinSF1     1460 non-null   int64  
 35  BsmtFinType2   1422 non-null   object 
 36  BsmtFinSF2     1460 non-null   int64  
 37  BsmtUnfSF      1460 non-null   int64  
 38  TotalBsmtSF    1460 non-null   int64  
 39  Heating        1460 non-null   object 
 40  HeatingQC      1460 non-null   object 
 41  CentralAir     1460 non-null   object 
 42  Electrical     1459 non-null   object 
 43  1stFlrSF       1460 non-null   int64  
 44  2ndFlrSF       1460 non-null   int64  
 45  LowQualFinSF   1460 non-null   int64  
 46  GrLivArea      1460 non-null   int64  
 47  BsmtFullBath   1460 non-null   int64  
 48  BsmtHalfBath   1460 non-null   int64  
 49  FullBath       1460 non-null   int64  
 50  HalfBath       1460 non-null   int64  
 51  BedroomAbvGr   1460 non-null   int64  
 52  KitchenAbvGr   1460 non-null   int64  
 53  KitchenQual    1460 non-null   object 
 54  TotRmsAbvGrd   1460 non-null   int64  
 55  Functional     1460 non-null   object 
 56  Fireplaces     1460 non-null   int64  
 57  FireplaceQu    770 non-null    object 
 58  GarageType     1379 non-null   object 
 59  GarageYrBlt    1379 non-null   float64
 60  GarageFinish   1379 non-null   object 
 61  GarageCars     1460 non-null   int64  
 62  GarageArea     1460 non-null   int64  
 63  GarageQual     1379 non-null   object 
 64  GarageCond     1379 non-null   object 
 65  PavedDrive     1460 non-null   object 
 66  WoodDeckSF     1460 non-null   int64  
 67  OpenPorchSF    1460 non-null   int64  
 68  EnclosedPorch  1460 non-null   int64  
 69  3SsnPorch      1460 non-null   int64  
 70  ScreenPorch    1460 non-null   int64  
 71  PoolArea       1460 non-null   int64  
 72  PoolQC         7 non-null      object 
 73  Fence          281 non-null    object 
 74  MiscFeature    54 non-null     object 
 75  MiscVal        1460 non-null   int64  
 76  MoSold         1460 non-null   int64  
 77  YrSold         1460 non-null   int64  
 78  SaleType       1460 non-null   object 
 79  SaleCondition  1460 non-null   object 
 80  SalePrice      1460 non-null   int64  
dtypes: float64(3), int64(35), object(43)
memory usage: 924.0+ KB

• 일부 결측치가 존재함을 확인할 수 있다.

2. 데이터 구성(데이터 전처리)

• setup() 함수를 통해 지정

reg = setup(data = train, 
             target = 'SalePrice',
             numeric_imputation = 'mean',
             categorical_features = ['MSZoning','Exterior1st','Exterior2nd','KitchenQual','Functional','SaleType',
                                     'Street','LotShape','LandContour','LotConfig','LandSlope','Neighborhood',   
                                     'Condition1','Condition2','BldgType','HouseStyle','RoofStyle','RoofMatl',    
                                     'MasVnrType','ExterQual','ExterCond','Foundation','BsmtQual','BsmtCond',   
                                     'BsmtExposure','BsmtFinType1','BsmtFinType2','Heating','HeatingQC','CentralAir',   
                                     'Electrical','GarageType','GarageFinish','GarageQual','GarageCond','PavedDrive',
                                     'SaleCondition']  , 
             ignore_features = ['Alley','PoolQC','MiscFeature','Fence','FireplaceQu','Utilities'],
             normalize = True)

	Description	Value
0	Session id	679
1	Target	SalePrice
2	Target type	Regression
3	Original data shape	(1460, 81)
4	Transformed data shape	(1460, 262)
5	Transformed train set shape	(1021, 262)
6	Transformed test set shape	(439, 262)
7	Ignore features	6
8	Ordinal features	2
9	Numeric features	37
10	Categorical features	37
11	Rows with missing values	100.0%
12	Preprocess	True
13	Imputation type	simple
14	Numeric imputation	mean
15	Categorical imputation	mode
16	Maximum one-hot encoding	25
17	Encoding method	None
18	Normalize	True
19	Normalize method	zscore
20	Fold Generator	KFold
21	Fold Number	10
22	CPU Jobs	-1
23	Use GPU	False
24	Log Experiment	False
25	Experiment Name	reg-default-name
26	USI	6ade

3. 모델 성능 비교

• models()를 통해 모델을 확인할 수 있음

  • setup을 한 후에만 가능

• compare_models()

  • models()에서 제공하는 모델들이나 scikit-learn에서 제공하는 모델을 별도로 선언한 이후에 입력한 모델들의 성능(MAE, MSE, RMSE, R^2, train time)등을 데이터프레임 형태로 제공

compare_models()

Initiated	. . . . . . . . . . . . . . . . . .	05:49:16
Status	. . . . . . . . . . . . . . . . . .	Fitting 10 Folds
Estimator	. . . . . . . . . . . . . . . . . .	Random Forest Regressor

	Model	MAE	MSE	RMSE	R2	RMSLE	MAPE	TT (Sec)
en	Elastic Net	18364.3952	1344589957.1260	35059.9474	0.7834	0.1449	0.1024	0.8530
par	Passive Aggressive Regressor	18134.1562	1374167341.7881	35373.6928	0.7773	0.1699	0.1033	2.4800
br	Bayesian Ridge	19102.7516	1509230751.8566	36789.4023	0.7551	0.1706	0.1097	0.9030
omp	Orthogonal Matching Pursuit	19565.0295	1604341893.6740	37530.4782	0.7370	0.1815	0.1121	1.2170
llar	Lasso Least Angle Regression	19787.4115	1620639261.1369	38028.6722	0.7366	0.1867	0.1157	0.9700
dt	Decision Tree Regressor	26873.8594	1656701282.2708	40601.1973	0.7338	0.2134	0.1554	0.9180
ridge	Ridge Regression	20390.6023	1675546296.8694	38612.7187	0.7280	0.2029	0.1197	1.1390
lasso	Lasso Regression	20348.5520	1679664973.5803	38597.0093	0.7271	0.2045	0.1196	3.0070
huber	Huber Regressor	19236.5019	1766235862.2651	38848.9916	0.7128	0.2052	0.1123	1.2150
knn	K Neighbors Regressor	27602.8084	2038986154.7901	44370.3270	0.6870	0.2092	0.1559	1.1220
lr	Linear Regression	489462244852579.7500	25087786038259630322198507421696.0000	3296273264514296.0000	-3909394621073952604160.0000	3.5936	4067501731.2203	4.4590
lar	Least Angle Regression	42521860416708132924698326635424220409273253888.0000	1844180973733323350448562275845971804997761547450196885667534503848224189678324096601303512252416.0000	429448190934811211097724065035493310127918809088.0000	-356872355765253379694331818216569495633078728307084395869623944858328491348458530144256.0000	49.8038	310381041719748206784084963210196848476160.0000	1.4570

Processing:   0%|          | 0/81 [00:00<?, ?it/s]

	Model	MAE	MSE	RMSE	R2	RMSLE	MAPE	TT (Sec)
gbr	Gradient Boosting Regressor	16668.8275	813316846.6913	27912.6374	0.8666	0.1348	0.0963	1.9690
rf	Random Forest Regressor	18261.1246	998294529.0476	31130.8689	0.8383	0.1493	0.1068	3.5390
lightgbm	Light Gradient Boosting Machine	17672.3315	1012800681.3536	31313.5997	0.8345	0.1418	0.0999	1.3330
xgboost	Extreme Gradient Boosting	19010.3780	1029452959.8392	31447.4011	0.8309	0.1482	0.1083	2.6500
et	Extra Trees Regressor	18237.5227	1037767632.2362	31830.1485	0.8297	0.1515	0.1073	3.0070
en	Elastic Net	18364.3952	1344589957.1260	35059.9474	0.7834	0.1449	0.1024	0.8530
par	Passive Aggressive Regressor	18134.1562	1374167341.7881	35373.6928	0.7773	0.1699	0.1033	2.4800
ada	AdaBoost Regressor	25297.1088	1483109549.4956	38170.5192	0.7603	0.2061	0.1631	1.6810
br	Bayesian Ridge	19102.7516	1509230751.8566	36789.4023	0.7551	0.1706	0.1097	0.9030
omp	Orthogonal Matching Pursuit	19565.0295	1604341893.6740	37530.4782	0.7370	0.1815	0.1121	1.2170
llar	Lasso Least Angle Regression	19787.4115	1620639261.1369	38028.6722	0.7366	0.1867	0.1157	0.9700
dt	Decision Tree Regressor	26873.8594	1656701282.2708	40601.1973	0.7338	0.2134	0.1554	0.9180
ridge	Ridge Regression	20390.6023	1675546296.8694	38612.7187	0.7280	0.2029	0.1197	1.1390
lasso	Lasso Regression	20348.5520	1679664973.5803	38597.0093	0.7271	0.2045	0.1196	3.0070
huber	Huber Regressor	19236.5019	1766235862.2651	38848.9916	0.7128	0.2052	0.1123	1.2150
knn	K Neighbors Regressor	27602.8084	2038986154.7901	44370.3270	0.6870	0.2092	0.1559	1.1220
dummy	Dummy Regressor	57764.4783	6442602524.5583	79947.6263	-0.0024	0.4042	0.3610	1.1870
lr	Linear Regression	489462244852579.7500	25087786038259630322198507421696.0000	3296273264514296.0000	-3909394621073952604160.0000	3.5936	4067501731.2203	4.4590
lar	Least Angle Regression	42521860416708132924698326635424220409273253888.0000	1844180973733323350448562275845971804997761547450196885667534503848224189678324096601303512252416.0000	429448190934811211097724065035493310127918809088.0000	-356872355765253379694331818216569495633078728307084395869623944858328491348458530144256.0000	49.8038	310381041719748206784084963210196848476160.0000	1.4570

GradientBoostingRegressor(random_state=679)

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

• LGBM의 성능이 가장 좋다.

4. 단일 모델 생성

• create_model()

  • 여러 모델이 아닌 하나의 모델에 대해서 setup()의 설정대로 학습을 진행하고 학습 결과 확인

  • 세부적으로 각 fold에 대한 성능을 제시

lgb = create_model('lightgbm')

Initiated	. . . . . . . . . . . . . . . . . .	05:55:18
Status	. . . . . . . . . . . . . . . . . .	Loading Dependencies
Estimator	. . . . . . . . . . . . . . . . . .	Compiling Library

Processing:   0%|          | 0/4 [00:00<?, ?it/s]

	MAE	MSE	RMSE	R2	RMSLE	MAPE
Fold
0	15442.1645	500820041.6930	22379.0090	0.9257	0.1161	0.0871
1	18244.9876	855920787.4039	29256.1239	0.8332	0.1369	0.1017
2	20166.5540	1435737692.1195	37891.1295	0.7530	0.1675	0.1193
3	17853.1263	1729976226.4035	41592.9829	0.6375	0.1523	0.1004
4	20165.9642	1283772957.2518	35829.7775	0.8314	0.1748	0.1221
5	16520.0521	686462662.0092	26200.4325	0.9053	0.1295	0.0917
6	17028.9203	786230258.6556	28039.7978	0.8479	0.1132	0.0887
7	16193.2443	1033335960.9131	32145.5434	0.8559	0.1365	0.0908
8	16631.2041	685488507.1789	26181.8354	0.8899	0.1640	0.1019
9	18477.0974	1130261719.9075	33619.3653	0.8654	0.1274	0.0954
Mean	17672.3315	1012800681.3536	31313.5997	0.8345	0.1418	0.0999
Std	1530.8375	365042550.9815	5679.7143	0.0797	0.0206	0.0116

• verbose 옵션

  • 함수 수행 시 발생하는 상세한 정보들을 표준 출력으로 자세히 내보낼 것인가를 결정

  • True의 경우 자세히 출력함

5. 모델 튜닝하기

• tune_model()

  • 입력한 모델에 대해서 hyper parameter tuning을 수행

tuned_lgb = tune_model(lgb)

	MAE	MSE	RMSE	R2	RMSLE	MAPE
Fold
0	16220.2312	672187894.3403	25926.5866	0.9003	0.1230	0.0895
1	18389.4504	957053892.6023	30936.2876	0.8135	0.1396	0.1032
2	18155.1805	1124713680.5994	33536.7512	0.8065	0.1665	0.1146
3	17656.8498	1042311405.3105	32284.8479	0.7816	0.1439	0.1019
4	18224.1143	1133540184.1908	33668.0885	0.8512	0.1765	0.1181
5	16493.6110	687283374.5380	26216.0900	0.9052	0.1354	0.0950
6	14976.2889	584667369.0686	24179.8960	0.8869	0.1024	0.0797
7	19611.3610	1654645129.8896	40677.3294	0.7693	0.1606	0.1084
8	16973.8506	744827085.8664	27291.5204	0.8803	0.1582	0.1068
9	20922.3215	1753138613.2070	41870.4981	0.7912	0.1580	0.1162
Mean	17762.3259	1035436862.9613	31658.7896	0.8386	0.1464	0.1033
Std	1629.3522	382505984.9453	5758.2901	0.0494	0.0211	0.0117

Processing:   0%|          | 0/7 [00:00<?, ?it/s]

Fitting 10 folds for each of 10 candidates, totalling 100 fits

6. 모델링 결과 해석

• interpret_model()

  • 모델을 해석

  • 훈련된 모델 객체와 플롯 유형을 문자열로 받음

  • 해석은 SHAP를 기반으로 구현되며 트리 기반 모델에서만 사용 가능함

cf) pycaret.classification, pycaret.regression 모듈에서만 사용 가능

cf) SHAP: SHapley Addictive exPlanations

!pip install shap

interpret_model(tuned_lgb)

predictions = predict_model(tuned_lgb, data = test)
sample['SalePrice'] = predictions['Label']
sample.to_csv('/content/drive/MyDrive/Colab Notebooks/ECC 48기 데과B/10주차/data/house_sample_submission.csv',index = False)
sample.head()