13wk-50: 아이스크림(이상치) / 자료분석(Autogluon)

Author

최규빈

Published

December 1, 2023

1. 강의영상

2. Imports

#!pip install autogluon.eda

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt 
import seaborn as sns
#---#"
from autogluon.tabular import TabularPredictor
import autogluon.eda.auto as auto
#---#
import warnings
warnings.filterwarnings('ignore')

3. Data

np.random.seed(43052)
temp = pd.read_csv('https://raw.githubusercontent.com/guebin/DV2022/master/posts/temp.csv').iloc[:100,3].to_numpy()
temp.sort()
ice_sales = 10 + temp * 0.5 + np.random.randn(100)
ice_sales[0] = 50
df_train = pd.DataFrame({'temp':temp,'ice_sales':ice_sales})
df_train

	temp	ice_sales
0	-4.1	50.000000
1	-3.7	9.234175
2	-3.0	9.642778
3	-1.3	9.657894
4	-0.5	9.987787
...	...	...
95	12.4	17.508688
96	13.4	17.105376
97	14.7	17.164930
98	15.0	18.555388
99	15.2	18.787014

100 rows × 2 columns

sns.scatterplot(df_train,x='temp',y='ice_sales')

4. 적합

# step1 -- pass 
# step2
predictr = TabularPredictor(label='ice_sales')
# step3
predictr.fit(df_train)
# step4
yhat = predictr.predict(df_train)

No path specified. Models will be saved in: "AutogluonModels/ag-20231201_104555/"
Beginning AutoGluon training ...
AutoGluon will save models to "AutogluonModels/ag-20231201_104555/"
AutoGluon Version:  0.8.2
Python Version:     3.10.13
Operating System:   Linux
Platform Machine:   x86_64
Platform Version:   #26~22.04.1-Ubuntu SMP PREEMPT_DYNAMIC Thu Jul 13 16:27:29 UTC 2
Disk Space Avail:   248.42 GB / 490.57 GB (50.6%)
Train Data Rows:    100
Train Data Columns: 1
Label Column: ice_sales
Preprocessing data ...
AutoGluon infers your prediction problem is: 'regression' (because dtype of label-column == float and many unique label-values observed).
    Label info (max, min, mean, stddev): (50.0, 8.273155164108418, 13.17881, 4.33788)
    If 'regression' is not the correct problem_type, please manually specify the problem_type parameter during predictor init (You may specify problem_type as one of: ['binary', 'multiclass', 'regression'])
Using Feature Generators to preprocess the data ...
Fitting AutoMLPipelineFeatureGenerator...
    Available Memory:                    126799.58 MB
    Train Data (Original)  Memory Usage: 0.0 MB (0.0% of available memory)
    Inferring data type of each feature based on column values. Set feature_metadata_in to manually specify special dtypes of the features.
    Stage 1 Generators:
        Fitting AsTypeFeatureGenerator...
    Stage 2 Generators:
        Fitting FillNaFeatureGenerator...
    Stage 3 Generators:
        Fitting IdentityFeatureGenerator...
    Stage 4 Generators:
        Fitting DropUniqueFeatureGenerator...
    Stage 5 Generators:
        Fitting DropDuplicatesFeatureGenerator...
    Types of features in original data (raw dtype, special dtypes):
        ('float', []) : 1 | ['temp']
    Types of features in processed data (raw dtype, special dtypes):
        ('float', []) : 1 | ['temp']
    0.0s = Fit runtime
    1 features in original data used to generate 1 features in processed data.
    Train Data (Processed) Memory Usage: 0.0 MB (0.0% of available memory)
Data preprocessing and feature engineering runtime = 0.02s ...
AutoGluon will gauge predictive performance using evaluation metric: 'root_mean_squared_error'
    This metric's sign has been flipped to adhere to being higher_is_better. The metric score can be multiplied by -1 to get the metric value.
    To change this, specify the eval_metric parameter of Predictor()
Automatically generating train/validation split with holdout_frac=0.2, Train Rows: 80, Val Rows: 20
User-specified model hyperparameters to be fit:
{
    'NN_TORCH': {},
    'GBM': [{'extra_trees': True, 'ag_args': {'name_suffix': 'XT'}}, {}, 'GBMLarge'],
    'CAT': {},
    'XGB': {},
    'FASTAI': {},
    'RF': [{'criterion': 'gini', 'ag_args': {'name_suffix': 'Gini', 'problem_types': ['binary', 'multiclass']}}, {'criterion': 'entropy', 'ag_args': {'name_suffix': 'Entr', 'problem_types': ['binary', 'multiclass']}}, {'criterion': 'squared_error', 'ag_args': {'name_suffix': 'MSE', 'problem_types': ['regression', 'quantile']}}],
    'XT': [{'criterion': 'gini', 'ag_args': {'name_suffix': 'Gini', 'problem_types': ['binary', 'multiclass']}}, {'criterion': 'entropy', 'ag_args': {'name_suffix': 'Entr', 'problem_types': ['binary', 'multiclass']}}, {'criterion': 'squared_error', 'ag_args': {'name_suffix': 'MSE', 'problem_types': ['regression', 'quantile']}}],
    'KNN': [{'weights': 'uniform', 'ag_args': {'name_suffix': 'Unif'}}, {'weights': 'distance', 'ag_args': {'name_suffix': 'Dist'}}],
}
Fitting 11 L1 models ...
Fitting model: KNeighborsUnif ...
    -2.2756  = Validation score   (-root_mean_squared_error)
    0.01s    = Training   runtime
    0.01s    = Validation runtime
Fitting model: KNeighborsDist ...
    -2.6883  = Validation score   (-root_mean_squared_error)
    0.01s    = Training   runtime
    0.01s    = Validation runtime
Fitting model: LightGBMXT ...
    -12.8059     = Validation score   (-root_mean_squared_error)
    0.47s    = Training   runtime
    0.0s     = Validation runtime
Fitting model: LightGBM ...
    -1.6843  = Validation score   (-root_mean_squared_error)
    0.17s    = Training   runtime
    0.0s     = Validation runtime
Fitting model: RandomForestMSE ...
    -2.562   = Validation score   (-root_mean_squared_error)
    0.27s    = Training   runtime
    0.02s    = Validation runtime
Fitting model: CatBoost ...
    -1.2192  = Validation score   (-root_mean_squared_error)
    0.25s    = Training   runtime
    0.0s     = Validation runtime
Fitting model: ExtraTreesMSE ...
    -1.8631  = Validation score   (-root_mean_squared_error)
    0.29s    = Training   runtime
    0.1s     = Validation runtime
Fitting model: NeuralNetFastAI ...
    -1.752   = Validation score   (-root_mean_squared_error)
    0.64s    = Training   runtime
    0.01s    = Validation runtime
Fitting model: XGBoost ...
    -1.3017  = Validation score   (-root_mean_squared_error)
    0.1s     = Training   runtime
    0.0s     = Validation runtime
Fitting model: NeuralNetTorch ...
    -1.1863  = Validation score   (-root_mean_squared_error)
    0.84s    = Training   runtime
    0.01s    = Validation runtime
Fitting model: LightGBMLarge ...
    -2.0798  = Validation score   (-root_mean_squared_error)
    0.19s    = Training   runtime
    0.0s     = Validation runtime
Fitting model: WeightedEnsemble_L2 ...
    -1.1737  = Validation score   (-root_mean_squared_error)
    0.22s    = Training   runtime
    0.0s     = Validation runtime
AutoGluon training complete, total runtime = 3.75s ... Best model: "WeightedEnsemble_L2"
TabularPredictor saved. To load, use: predictor = TabularPredictor.load("AutogluonModels/ag-20231201_104555/")

[1000]  valid_set's rmse: 1.69432

sns.scatterplot(df_train,x='temp',y='ice_sales',alpha=0.2)
sns.lineplot(df_train,x='temp',y=yhat)

5. 해석 및 시각화

A. y의 분포, (X,y)의 관계 시각화

auto.target_analysis(
    train_data=df_train,
    label='ice_sales',
    fit_distributions=False
)

Target variable analysis

	count	mean	std	min	25%	50%	75%	max	dtypes	unique	missing_count	missing_ratio	raw_type	special_types
ice_sales	100	13.178805	4.337878	8.273155	11.296645	12.856589	14.294614	50.0	float64	100			float

Target variable correlations

train_data - spearman correlation matrix; focus: absolute correlation for ice_sales >= 0.5

Feature interaction between temp/ice_sales in train_data

B. 중요한 설명변수?

pass # 설명변수가 하나라서..

C. 관측치별 해석

df_train.iloc[[0]]

	temp	ice_sales
0	-4.1	50.0

predictr.predict(df_train.iloc[[0]])

0    18.230947
Name: ice_sales, dtype: float32

auto.explain_rows(
    train_data=df_train,
    model=predictr,
    rows=df_train.iloc[[0]],
    display_rows= True,
    plot='waterfall'
)

	temp	ice_sales
0	-4.1	50.0