[Classification] Bank Marketing

Main Conclusion

Increasing the number of contacts with clients during the campaign does not necessarily result in a higher probability of positive contributions to the "deposit." Therefore, as a strategy to enhance marketing effectiveness, it is considered desirable to minimize contact with customers while increasing the contact duration. Particularly, marketing outcomes do not seem to be significantly influenced by the customer's age. It appears more favorable to focus on customers who were successful in previous marketing campaigns and students.

 

 

---

Summary for Results by Analysis Procedures

Step01. Data Extraction

• Missing Value & Duplication Inspection: It has been confirmed that there are no data missing or duplicates, and the the training dataset consists of 45,211 instances while the test dataset 4,521 instances.

Step02. Exploratory Data Analysis

• Descriptive Statistics: The datasets consists of a total of 17 columns, including 10 columns of string type and 6 columns of integer or floating-point type.
• Exploration of independence between features: It has been confirmed that there are no explainatory variables (16 columns) that maintain independence.
• Exploration of association with target feature: Based on decision tree and association analysis, it is inferred that the attributes "duration", "balance", "poutcome", and "month" are the major factors that determine the "deposit"(target feature).

Step03. Linear Relationship Analysis

• Regression Analysis: It has been observed that the attributes "default", "marital", "load", "education", and "age" exhibit a strong multicollinearity. As a result, the attribute that contributes most positively to the "deposit" is "duration". Specifically, when the job is "student" or the previous marketing campaign was successful, there is a higher probability of positive contribution to the "deposit." On the other hand, the attribute "campaign" has a negative contribution. When the job is "student" and the education level is "tertiary," the probability of positive contribution to the "deposit" is low.
• Covariance Analysis: It has been observed that the combinations of explanatory variables with positive correlation are ("job", "education"), ("pdays", "previous"), ("pdays", "poutcome"), and ("previous", "poutcome"). On the other hand, the combination of ("age", "marital") has been identified to have a negative correlation. Futhermore, it has been determined that the target variable has a positive correlation with "duration", "poutcome", and "month".
• Exploratory Factor Analysis: It has been observed that "pdays", "previous", and "poutcome" share a common factor related to previous campaigns. Additionally, it has been confimed that "age" and "marital" also share customer demographic attributes. Moreover, "age", "job", "education", "housing", "contact", and "month" have been found to interact synergistically with the "deposit".

Step04. Predictive Modeling and Evaluation

• Effect validation by data preprocessing scenarioes: Based on the recall score and the LDA model, four main preprocessing approaches were explored. It appears that the prior effect, upsampling, addressing modeling assumptions approaches have shown positive effects. However, outliers handling seems to have demonstrated no significant impact on the results. 
• Validation baseline for training models: After conducting two rounds of testing on the sampled data, it has been observed that the baseline model yielded low recall scores. In order to improve the recall score, various modeling approaches are being explored. LDA, KNN, and GBC models are considered effective for enhancing performace.
• AI Model Evaluation: Great job on improving the recall score of the KNN baseline model from 0.3647 to 0.9674. However, it's important to note that there is a trade-off effect as the precision decreased from 0.5460 to 0.4448.

Step05. Summary for Results by Analysis Procedures

Step06. Main Conclusion

 

 

---

Predictive Modeling and Evaluation

AI Model Evaluation

Final test result

model	total	TP	TN	FP	FN	accuracy	precision	recall	f1
logisticregression	4521	374	3500	500	147	0.856890	0.427918	0.717850	0.536201
lineardiscriminantanalysis	4521	359	3571	429	162	0.869277	0.455584	0.689060	0.548510
svc	4521	361	3200	800	160	0.787658	0.310939	0.692898	0.429251
kneighborsclassifier	4521	504	3371	629	17	0.857111	0.444837	0.967370	0.609432
extratreeclassifier	4521	492	1959	2041	29	0.542137	0.194236	0.944338	0.322200
gradientboostingclassifier	4521	520	637	3363	1	0.255917	0.133917	0.998081	0.236149

cross-validation for full dataset: 2-fold & 2-repeated

 

 

 

 

 

Validation baseline for training models

Modeling

probabilistic generative models: GNB, LDA, QDA, ...

main objective: probabilistic interpretation for conditional distribution of features on data

• selected validation models: 'linear discriminative analysis' model
  
  • checking point: conditional independence viloation of the distribution of individual classes

probabilistic discriminative models: KNN, DT, RF, Ensembles, Logit, SVM, NN, ...

main objective: probabilistic interpretation for information quantity(i.e. information gain) of each features on data

• selected validation models: 'k-nearst neighbors', 'extra tree', 'gradient boosting ensemble'
  • checking point: the hard or soft decision boundary between classes
• selected validation models: 'logistic regression', 'support vector machine' model
  • checking point: feature independence viloation

 

Hyper-parameter ranges to prevent overfitting during learning

Model	hyper-parameter1	hyper-parameter2	hyper-parameter3
Logistic	C: [0.001, 0.005, 0.007]
LDA	priors: [(0.1, 0.9), (0.2, 0.8), (0.3, 0.7)]
SVC	C: [0.01, 0.05, 0.07]
KNN	n_neighbors: [20, 30]	leaf_size: [30, 50, 100]
ETC	min_impurity_decrease: [0.01, 0.05, 0.1]	max_depth: [10, 20, 30]
GBC	min_impurity_decrease: [0.01, 0.05, 0.1]	n_estimators: [10, 30, 50]	subsample: [0.7, 0.8, 1]

 

Transformers for data preprocessing

Objective : preprocessing	Normality and decision boundry secure	Numericalization	Feature selection	Dimensionality reduction	Feature diversification
Model	Powertransformer	OnehotEncoder	SelectPercentile	PCA	SplineTransformer
Logistic	O	X	O	X	O
LDA	O	X	O	X	O
SVC	O	O	O	X	O
KNN	O	O	O	X	O
ETC	O	O	O	X	O
GBC	O	O	O	X	O

 

 

Validation Result

2nd test result

preprocessing for stratified sampling (frac=.1 & 3 repeated * 5 fold)

• upsampling SMOTE
• custom preprocessing for continous feature

model	total	TP	TN	FP	FN	accuracy	precision	recall	f1
logistic	4521	390	3469	531	131	0.8535	0.4234	0.7485	0.5409
LDA	4521	354	3551	449	167	0.8637	0.4408	0.6794	0.5347
SVC	4521	350	3241	759	171	0.7942	0.3156	0.6717	0.4294
KNN	4521	415	3041	959	106	0.7644	0.3020	0.7965	0.4379
ETC	4521	327	2507	1493	194	0.6268	0.1796	0.6276	0.2793
GBC	4521	520	677	3323	1	0.2647	0.1353	0.9980	0.2383

(Validation for Train Dataset) Scenario 1~6: Logistic, LDA, SVC, KNN, Extra-Tree, Gradient Boosting Ensemble

 

1st test result

baseline for stratified sampling (frac=.1 & 3 repeated * 5 fold)

model	total	TP	TN	FP	FN	accuracy	precision	recall	f1
Logistic	4521	142	3922	78	379	0.8989	0.6455	0.2726	0.3833
LDA	4521	198	3879	121	323	0.9018	0.6207	0.3800	0.4714
SVC	4521	0	4000	0	521	0.8848	0.0000	0.0000	0.0000
KNN	4521	190	3842	158	331	0.8918	0.5460	0.3647	0.4374
ETC	4521	98	3953	47	423	0.8960	0.6759	0.1881	0.2943
GBC	4521	221	3865	135	300	0.9038	0.6208	0.4242	0.5040

(Validation for Train Dataset) Scenario 1~6: Logistic, LDA, SVC, KNN, Extra-Tree, Gradient Boosting Ensemble

 

 

Effect validation by data preprocessing scenarioes

First, prior effect, this involves addressing the influence or prior information or biases in the data. Second, upsampling for class imbalance, to tackle class imbalance, upsampling techniques were applied to increase the representation of the minority class. Third, effects of addressing modeling assumptions, this includes addressing assumptions such as normality, standardization, and normalization to meet the modeling requirements. Lastly, outlier handling, the effect of outlier handling was examined, which involves identifying and dealing with data points that deviate significantly from the overall pattern. These four preprocessing approaches were evaluated in terms of their impact on the performance, specifically with regard to the recall score and the LDA model.

 

 

Cost-sensitive priors

Priors: effective

	test_recall
param_priors
(0.1, 0.9)	0.9210
(0.2, 0.8)	0.8801
(0.3, 0.7)	0.8427
(0.4, 0.6)	0.7973
(0.5, 0.5)	0.7512
(0.6, 0.4)	0.6948
(0.7, 0.3)	0.6347
(0.8, 0.2)	0.5752
(0.9, 0.1)	0.4925

	Source	SS	DF	MS	F	p-unc	np2
0	param_priors	0.8366	8	0.1046	284.0343	0.0	0.9844
1	Within	0.0133	36	0.0004	NaN	NaN	NaN

	A(no, yes)	B(no, yes)	mean(A)	mean(B)	diff	se	T	p-tukey	hedges
0	(0.1, 0.9)	(0.2, 0.8)	0.9210	0.8801	0.0408	0.0121	3.3652	0.0424	4.8871
1	(0.1, 0.9)	(0.3, 0.7)	0.9210	0.8427	0.0783	0.0121	6.4498	0.0000	4.9797
2	(0.1, 0.9)	(0.4, 0.6)	0.9210	0.7973	0.1237	0.0121	10.1891	0.0000	6.1375
3	(0.1, 0.9)	(0.5, 0.5)	0.9210	0.7512	0.1698	0.0121	13.9904	0.0000	8.8321
4	(0.1, 0.9)	(0.6, 0.4)	0.9210	0.6948	0.2261	0.0121	18.6331	0.0000	13.1020
5	(0.1, 0.9)	(0.7, 0.3)	0.9210	0.6347	0.2863	0.0121	23.5874	0.0000	15.2181
6	(0.1, 0.9)	(0.8, 0.2)	0.9210	0.5752	0.3458	0.0121	28.4951	0.0000	23.8565
7	(0.1, 0.9)	(0.9, 0.1)	0.9210	0.4925	0.4284	0.0121	35.3035	0.0000	26.6182
8	(0.2, 0.8)	(0.3, 0.7)	0.8801	0.8427	0.0374	0.0121	3.0846	0.0819	2.3003
9	(0.2, 0.8)	(0.4, 0.6)	0.8801	0.7973	0.0828	0.0121	6.8238	0.0000	4.0234
10	(0.2, 0.8)	(0.5, 0.5)	0.8801	0.7512	0.1289	0.0121	10.6252	0.0000	6.5521
11	(0.2, 0.8)	(0.6, 0.4)	0.8801	0.6948	0.1853	0.0121	15.2679	0.0000	10.4294
12	(0.2, 0.8)	(0.7, 0.3)	0.8801	0.6347	0.2454	0.0121	20.2222	0.0000	12.7315
13	(0.2, 0.8)	(0.8, 0.2)	0.8801	0.5752	0.3050	0.0121	25.1298	0.0000	20.2030
14	(0.2, 0.8)	(0.9, 0.1)	0.8801	0.4925	0.3876	0.0121	31.9383	0.0000	23.2960
15	(0.3, 0.7)	(0.4, 0.6)	0.8427	0.7973	0.0454	0.0121	3.7392	0.0164	1.8512
16	(0.3, 0.7)	(0.5, 0.5)	0.8427	0.7512	0.0915	0.0121	7.5406	0.0000	3.8515
17	(0.3, 0.7)	(0.6, 0.4)	0.8427	0.6948	0.1479	0.0121	12.1833	0.0000	6.6599
18	(0.3, 0.7)	(0.7, 0.3)	0.8427	0.6347	0.2080	0.0121	17.1376	0.0000	8.8779
19	(0.3, 0.7)	(0.8, 0.2)	0.8427	0.5752	0.2675	0.0121	22.0452	0.0000	13.2921
20	(0.3, 0.7)	(0.9, 0.1)	0.8427	0.4925	0.3502	0.0121	28.8537	0.0000	16.4328
21	(0.4, 0.6)	(0.5, 0.5)	0.7973	0.7512	0.0461	0.0121	3.8013	0.0140	1.7153
22	(0.4, 0.6)	(0.6, 0.4)	0.7973	0.6948	0.1025	0.0121	8.4440	0.0000	4.0142
23	(0.4, 0.6)	(0.7, 0.3)	0.7973	0.6347	0.1626	0.0121	13.3984	0.0000	6.1125
24	(0.4, 0.6)	(0.8, 0.2)	0.7973	0.5752	0.2222	0.0121	18.3060	0.0000	9.3551
25	(0.4, 0.6)	(0.9, 0.1)	0.7973	0.4925	0.3048	0.0121	25.1145	0.0000	12.3113
26	(0.5, 0.5)	(0.6, 0.4)	0.7512	0.6948	0.0563	0.0121	4.6427	0.0013	2.2713
27	(0.5, 0.5)	(0.7, 0.3)	0.7512	0.6347	0.1165	0.0121	9.5970	0.0000	4.4953
28	(0.5, 0.5)	(0.8, 0.2)	0.7512	0.5752	0.1760	0.0121	14.5047	0.0000	7.6636
29	(0.5, 0.5)	(0.9, 0.1)	0.7512	0.4925	0.2587	0.0121	21.3131	0.0000	10.7722
30	(0.6, 0.4)	(0.7, 0.3)	0.6948	0.6347	0.0601	0.0121	4.9543	0.0005	2.4554
31	(0.6, 0.4)	(0.8, 0.2)	0.6948	0.5752	0.1197	0.0121	9.8620	0.0000	5.6052
32	(0.6, 0.4)	(0.9, 0.1)	0.6948	0.4925	0.2023	0.0121	16.6704	0.0000	9.0039
33	(0.7, 0.3)	(0.8, 0.2)	0.6347	0.5752	0.0596	0.0121	4.9076	0.0006	2.6325
34	(0.7, 0.3)	(0.9, 0.1)	0.6347	0.4925	0.1422	0.0121	11.7161	0.0000	6.0041
35	(0.8, 0.2)	(0.9, 0.1)	0.5752	0.4925	0.0826	0.0121	6.8085	0.0000	4.0457

Cost-sensitive sampling for target class balancing

• Under sampling(X)
• Over sampling(O) : a little bit effective
• Combined sampling(X)

	test_recall
sampling_strategy
0.0	0.303053
0.5	0.805162
0.8	0.866887
0.9	0.882855
1.0	0.893989

	F Value	Num DF	Den DF	Pr > F
sampling_strategy	15.723825	4.0	16.0	0.000021

		stat	pval	pval_corr	reject
group1	group2
0.0	0.5	-2.7108	0.0266	0.2663	False
	0.8	-3.2577	0.0116	0.1157	False
	0.9	-3.5244	0.0078	0.078	False
	1.0	-3.7396	0.0057	0.0571	False
0.5	0.8	-0.3519	0.734	1.0	False
	0.9	-0.4653	0.6541	1.0	False
	1.0	-0.5531	0.5953	1.0	False
0.8	0.9	-0.1041	0.9197	1.0	False
	1.0	-0.1851	0.8577	1.0	False
0.9	1.0	-0.0818	0.9368	1.0	False

Data transformation

• Linear independence of the features: Model Assumption
  • Nonlinear transform: Normality; GNB, LDA, QDA
  • Linear transform: Standard Scaling (Z-Transform) for LDA
  • Constraint: Normalization for LDA, QDA 
• Whitening Distribution Outlier
  • Robust Scaling / Minmax Scaling, Maxabs Scaling

Linearity effect: effective

	test_recall
treatment
_	0.3621
_H	0.3621
_HV	0.3849
_N	0.2516
_NH	0.2516
_NHV	0.1976
_NV	0.0000
_V	0.0737

	sum_sq	df	F	PR(>F)
C(normality)	0.1452	1.0	9.7369	0.0038
C(heteroscedasticity)	0.1618	1.0	10.8528	0.0024
C(vectorspace)	0.2039	1.0	13.6769	0.0008
C(normality):C(heteroscedasticity)	0.0081	1.0	0.5414	0.4672
C(heteroscedasticity):C(vectorspace)	0.1618	1.0	10.8528	0.0024
C(normality):C(vectorspace)	0.0010	1.0	0.0680	0.7960
C(normality):C(heteroscedasticity):C(vectorspace)	0.0081	1.0	0.5414	0.4672
Residual	0.4770	32.0	NaN	NaN

		meandiff	p-adj	lower	upper	reject
group1	group2
_N	_NV	-0.2516	0.0478	-0.5018	-0.0015	True
_NH	_NV	-0.2516	0.0478	-0.5018	-0.0015	True
_	_V	-0.2883	0.0149	-0.5385	-0.0382	True
_H	_V	-0.2883	0.0149	-0.5385	-0.0382	True
_HV	_V	-0.3112	0.0069	-0.5614	-0.0611	True
_	_NV	-0.3621	0.0011	-0.6122	-0.1119	True
_H	_NV	-0.3621	0.0011	-0.6122	-0.1119	True
_HV	_NV	-0.3849	0.0005	-0.6351	-0.1348	True
_NV	_V	0.0737	0.9776	-0.1764	0.3239	False
_	_HV	0.0229	1.0	-0.2273	0.273	False
_H	_HV	0.0229	1.0	-0.2273	0.273	False
_	_H	0.0	1.0	-0.2501	0.2501	False
_N	_NH	0.0	1.0	-0.2501	0.2501	False
	_NHV	-0.0541	0.9964	-0.3042	0.1961	False
_NH	_NHV	-0.0541	0.9964	-0.3042	0.1961	False
_	_N	-0.1104	0.8367	-0.3606	0.1397	False
	_NH	-0.1104	0.8367	-0.3606	0.1397	False
_H	_N	-0.1104	0.8367	-0.3606	0.1397	False
	_NH	-0.1104	0.8367	-0.3606	0.1397	False
_NHV	_V	-0.1238	0.7446	-0.374	0.1263	False
_HV	_N	-0.1333	0.671	-0.3834	0.1168	False
	_NH	-0.1333	0.671	-0.3834	0.1168	False
_	_NHV	-0.1645	0.4184	-0.4146	0.0857	False
_H	_NHV	-0.1645	0.4184	-0.4146	0.0857	False
_N	_V	-0.1779	0.3224	-0.4281	0.0722	False
_NH	_V	-0.1779	0.3224	-0.4281	0.0722	False
_HV	_NHV	-0.1874	0.2634	-0.4375	0.0628	False
_NHV	_NV	-0.1976	0.2084	-0.4477	0.0526	False

Outlier effect: non-effective

		test_recall
scaler	contamination
A	0.00	0.260528
	0.01	0.260528
	0.05	0.260528
	0.10	0.260528
	0.20	0.260528
	0.50	0.260528
M	0.00	0.260528
	0.01	0.260528
	0.05	0.260528
	0.10	0.260528
	0.20	0.260528
	0.50	0.260528
R	0.00	0.279438
	0.01	0.279438
	0.05	0.279438
	0.10	0.279438
	0.20	0.279438
	0.50	0.279438

	sum_sq	df	F	PR(>F)
C(scaler)	7.151514e-03	2.0	1.613890e-01	0.851268
C(contamination)	2.223554e-32	5.0	2.007168e-31	1.000000
C(scaler):C(contamination)	1.617393e-31	10.0	7.299979e-31	1.000000
Residual	1.595242e+00	72.0	NaN	NaN

		meandiff	p-adj	lower	upper	reject
group1	group2
A0.0	R0.0	0.0189	1.0	-0.3217	0.3595	False
	R0.01	0.0189	1.0	-0.3217	0.3595	False
	R0.05	0.0189	1.0	-0.3217	0.3595	False
	R0.1	0.0189	1.0	-0.3217	0.3595	False
	R0.2	0.0189	1.0	-0.3217	0.3595	False
	R0.5	0.0189	1.0	-0.3217	0.3595	False
A0.01	R0.0	0.0189	1.0	-0.3217	0.3595	False
	R0.01	0.0189	1.0	-0.3217	0.3595	False
	R0.05	0.0189	1.0	-0.3217	0.3595	False
	R0.1	0.0189	1.0	-0.3217	0.3595	False
	R0.2	0.0189	1.0	-0.3217	0.3595	False
	R0.5	0.0189	1.0	-0.3217	0.3595	False
A0.05	R0.0	0.0189	1.0	-0.3217	0.3595	False
	R0.01	0.0189	1.0	-0.3217	0.3595	False
	R0.05	0.0189	1.0	-0.3217	0.3595	False
	R0.1	0.0189	1.0	-0.3217	0.3595	False
	R0.2	0.0189	1.0	-0.3217	0.3595	False
	R0.5	0.0189	1.0	-0.3217	0.3595	False
A0.1	R0.0	0.0189	1.0	-0.3217	0.3595	False
	R0.01	0.0189	1.0	-0.3217	0.3595	False
	R0.05	0.0189	1.0	-0.3217	0.3595	False
	R0.1	0.0189	1.0	-0.3217	0.3595	False
	R0.2	0.0189	1.0	-0.3217	0.3595	False
	R0.5	0.0189	1.0	-0.3217	0.3595	False
A0.2	R0.0	0.0189	1.0	-0.3217	0.3595	False
	R0.01	0.0189	1.0	-0.3217	0.3595	False
	R0.05	0.0189	1.0	-0.3217	0.3595	False
	R0.1	0.0189	1.0	-0.3217	0.3595	False
	R0.2	0.0189	1.0	-0.3217	0.3595	False
	R0.5	0.0189	1.0	-0.3217	0.3595	False
A0.5	R0.0	0.0189	1.0	-0.3217	0.3595	False
	R0.01	0.0189	1.0	-0.3217	0.3595	False
	R0.05	0.0189	1.0	-0.3217	0.3595	False
	R0.1	0.0189	1.0	-0.3217	0.3595	False
	R0.2	0.0189	1.0	-0.3217	0.3595	False
	R0.5	0.0189	1.0	-0.3217	0.3595	False
M0.0	R0.0	0.0189	1.0	-0.3217	0.3595	False
	R0.01	0.0189	1.0	-0.3217	0.3595	False
	R0.05	0.0189	1.0	-0.3217	0.3595	False
	R0.1	0.0189	1.0	-0.3217	0.3595	False
	R0.2	0.0189	1.0	-0.3217	0.3595	False
	R0.5	0.0189	1.0	-0.3217	0.3595	False
M0.01	R0.0	0.0189	1.0	-0.3217	0.3595	False
	R0.01	0.0189	1.0	-0.3217	0.3595	False
	R0.05	0.0189	1.0	-0.3217	0.3595	False
	R0.1	0.0189	1.0	-0.3217	0.3595	False
	R0.2	0.0189	1.0	-0.3217	0.3595	False
	R0.5	0.0189	1.0	-0.3217	0.3595	False
M0.05	R0.0	0.0189	1.0	-0.3217	0.3595	False
	R0.01	0.0189	1.0	-0.3217	0.3595	False
	R0.05	0.0189	1.0	-0.3217	0.3595	False
	R0.1	0.0189	1.0	-0.3217	0.3595	False
	R0.2	0.0189	1.0	-0.3217	0.3595	False
	R0.5	0.0189	1.0	-0.3217	0.3595	False
M0.1	R0.0	0.0189	1.0	-0.3217	0.3595	False
	R0.01	0.0189	1.0	-0.3217	0.3595	False
	R0.05	0.0189	1.0	-0.3217	0.3595	False
	R0.1	0.0189	1.0	-0.3217	0.3595	False
	R0.2	0.0189	1.0	-0.3217	0.3595	False
	R0.5	0.0189	1.0	-0.3217	0.3595	False
M0.2	R0.0	0.0189	1.0	-0.3217	0.3595	False
	R0.01	0.0189	1.0	-0.3217	0.3595	False
	R0.05	0.0189	1.0	-0.3217	0.3595	False
	R0.1	0.0189	1.0	-0.3217	0.3595	False
	R0.2	0.0189	1.0	-0.3217	0.3595	False
	R0.5	0.0189	1.0	-0.3217	0.3595	False
M0.5	R0.0	0.0189	1.0	-0.3217	0.3595	False
	R0.01	0.0189	1.0	-0.3217	0.3595	False
	R0.05	0.0189	1.0	-0.3217	0.3595	False
	R0.1	0.0189	1.0	-0.3217	0.3595	False
	R0.2	0.0189	1.0	-0.3217	0.3595	False
	R0.5	0.0189	1.0	-0.3217	0.3595	False
A0.0	A0.01	0.0	1.0	-0.3406	0.3406	False
	A0.05	0.0	1.0	-0.3406	0.3406	False
	A0.1	0.0	1.0	-0.3406	0.3406	False
	A0.2	0.0	1.0	-0.3406	0.3406	False
	A0.5	0.0	1.0	-0.3406	0.3406	False
	M0.0	0.0	1.0	-0.3406	0.3406	False
	M0.01	0.0	1.0	-0.3406	0.3406	False
	M0.05	0.0	1.0	-0.3406	0.3406	False
	M0.1	0.0	1.0	-0.3406	0.3406	False
	M0.2	0.0	1.0	-0.3406	0.3406	False
	M0.5	0.0	1.0	-0.3406	0.3406	False
A0.01	A0.05	0.0	1.0	-0.3406	0.3406	False
	A0.1	0.0	1.0	-0.3406	0.3406	False
	A0.2	0.0	1.0	-0.3406	0.3406	False
	A0.5	0.0	1.0	-0.3406	0.3406	False
	M0.0	0.0	1.0	-0.3406	0.3406	False
	M0.01	0.0	1.0	-0.3406	0.3406	False
	M0.05	0.0	1.0	-0.3406	0.3406	False
	M0.1	0.0	1.0	-0.3406	0.3406	False
	M0.2	0.0	1.0	-0.3406	0.3406	False
	M0.5	0.0	1.0	-0.3406	0.3406	False
A0.05	A0.1	0.0	1.0	-0.3406	0.3406	False
	A0.2	0.0	1.0	-0.3406	0.3406	False
	A0.5	0.0	1.0	-0.3406	0.3406	False
	M0.0	0.0	1.0	-0.3406	0.3406	False
	M0.01	0.0	1.0	-0.3406	0.3406	False
	M0.05	0.0	1.0	-0.3406	0.3406	False
	M0.1	0.0	1.0	-0.3406	0.3406	False
	M0.2	0.0	1.0	-0.3406	0.3406	False
	M0.5	0.0	1.0	-0.3406	0.3406	False
A0.1	A0.2	0.0	1.0	-0.3406	0.3406	False
	A0.5	0.0	1.0	-0.3406	0.3406	False
	M0.0	0.0	1.0	-0.3406	0.3406	False
	M0.01	0.0	1.0	-0.3406	0.3406	False
	M0.05	0.0	1.0	-0.3406	0.3406	False
	M0.1	0.0	1.0	-0.3406	0.3406	False
	M0.2	0.0	1.0	-0.3406	0.3406	False
	M0.5	0.0	1.0	-0.3406	0.3406	False
A0.2	A0.5	0.0	1.0	-0.3406	0.3406	False
	M0.0	0.0	1.0	-0.3406	0.3406	False
	M0.01	0.0	1.0	-0.3406	0.3406	False
	M0.05	0.0	1.0	-0.3406	0.3406	False
	M0.1	0.0	1.0	-0.3406	0.3406	False
	M0.2	0.0	1.0	-0.3406	0.3406	False
	M0.5	0.0	1.0	-0.3406	0.3406	False
A0.5	M0.0	0.0	1.0	-0.3406	0.3406	False
	M0.01	0.0	1.0	-0.3406	0.3406	False
	M0.05	0.0	1.0	-0.3406	0.3406	False
	M0.1	0.0	1.0	-0.3406	0.3406	False
	M0.2	0.0	1.0	-0.3406	0.3406	False
	M0.5	0.0	1.0	-0.3406	0.3406	False
M0.0	M0.01	0.0	1.0	-0.3406	0.3406	False
	M0.05	0.0	1.0	-0.3406	0.3406	False
	M0.1	0.0	1.0	-0.3406	0.3406	False
	M0.2	0.0	1.0	-0.3406	0.3406	False
	M0.5	0.0	1.0	-0.3406	0.3406	False
M0.01	M0.05	0.0	1.0	-0.3406	0.3406	False
	M0.1	0.0	1.0	-0.3406	0.3406	False
	M0.2	0.0	1.0	-0.3406	0.3406	False
	M0.5	0.0	1.0	-0.3406	0.3406	False
M0.05	M0.1	0.0	1.0	-0.3406	0.3406	False
	M0.2	0.0	1.0	-0.3406	0.3406	False
	M0.5	0.0	1.0	-0.3406	0.3406	False
M0.1	M0.2	0.0	1.0	-0.3406	0.3406	False
	M0.5	0.0	1.0	-0.3406	0.3406	False
M0.2	M0.5	0.0	1.0	-0.3406	0.3406	False
R0.0	R0.01	0.0	1.0	-0.3406	0.3406	False
	R0.05	0.0	1.0	-0.3406	0.3406	False
	R0.1	0.0	1.0	-0.3406	0.3406	False
	R0.2	0.0	1.0	-0.3406	0.3406	False
	R0.5	0.0	1.0	-0.3406	0.3406	False
R0.01	R0.05	0.0	1.0	-0.3406	0.3406	False
	R0.1	0.0	1.0	-0.3406	0.3406	False
	R0.2	0.0	1.0	-0.3406	0.3406	False
	R0.5	0.0	1.0	-0.3406	0.3406	False
R0.05	R0.1	0.0	1.0	-0.3406	0.3406	False
	R0.2	0.0	1.0	-0.3406	0.3406	False
	R0.5	0.0	1.0	-0.3406	0.3406	False
R0.1	R0.2	0.0	1.0	-0.3406	0.3406	False
	R0.5	0.0	1.0	-0.3406	0.3406	False
R0.2	R0.5	0.0	1.0	-0.3406	0.3406	False

 

 

 

 

---

Linear relationship analysis

Regression Analysis

(Note) Regression analysis has been conducted with one-hot encoding for categorical variables. 

 

 

Logit analysis summary table

• representative positive effect factors on deposit: C(poutcome)[T.success], C(month)[T.mar], C(job)[T.student], duration
• representative negative effect factors on deposit: C(contact)[unknown], C(contact)[telephone], C(contact)[cellular], C(month)[T.jan], campaign

with feature interaction

	centering							standardizing
	feature	coef	std err	z	P>|z|	[0.025	0.975]	feature	coef	std err	z	P>|z|	[0.025	0.975]
1	C(contact)[cellular]	-1.5389	0.136	-11.350	0.000	-1.805	-1.273	C(contact)[cellular]	-1.5389	0.136	-11.350	0.000	-1.805	-1.273
2	C(contact)[telephone]	-1.7394	0.153	-11.356	0.000	-2.040	-1.439	C(contact)[telephone]	-1.7394	0.153	-11.356	0.000	-2.040	-1.439
3	C(contact)[unknown]	-3.2081	0.157	-20.396	0.000	-3.516	-2.900	C(contact)[unknown]	-3.2081	0.157	-20.396	0.000	-3.516	-2.900
4	C(housing)[T.yes]	-0.6947	0.044	-15.956	0.000	-0.780	-0.609	C(housing)[T.yes]	-0.6947	0.044	-15.956	0.000	-0.780	-0.609
5	C(job)[T.blue-collar]	-0.4229	0.070	-6.069	0.000	-0.559	-0.286	C(job)[T.blue-collar]	-0.4229	0.070	-6.069	0.000	-0.559	-0.286
6	C(job)[T.entrepreneur]	-0.3586	0.123	-2.912	0.004	-0.600	-0.117	C(job)[T.entrepreneur]	-0.3586	0.123	-2.912	0.004	-0.600	-0.117
7	C(job)[T.housemaid]	-0.6010	0.133	-4.527	0.000	-0.861	-0.341	C(job)[T.housemaid]	-0.6010	0.133	-4.527	0.000	-0.861	-0.341
8	C(job)[T.management]	-0.0188	0.064	-0.295	0.768	-0.144	0.106	C(job)[T.management]	-0.0188	0.064	-0.295	0.768	-0.144	0.106
9	C(job)[T.retired]	0.1480	0.084	1.762	0.078	-0.017	0.313	C(job)[T.retired]	0.1480	0.084	1.762	0.078	-0.017	0.313
10	C(job)[T.self-employed]	-0.2023	0.109	-1.864	0.062	-0.415	0.010	C(job)[T.self-employed]	-0.2023	0.109	-1.864	0.062	-0.415	0.010
11	C(job)[T.services]	-0.2544	0.084	-3.037	0.002	-0.419	-0.090	C(job)[T.services]	-0.2544	0.084	-3.037	0.002	-0.419	-0.090
12	C(job)[T.student]	0.5845	0.104	5.638	0.000	0.381	0.788	C(job)[T.student]	0.5845	0.104	5.638	0.000	0.381	0.788
13	C(job)[T.technician]	-0.1352	0.068	-1.981	0.048	-0.269	-0.001	C(job)[T.technician]	-0.1352	0.068	-1.981	0.048	-0.269	-0.001
14	C(job)[T.unemployed]	-0.1438	0.110	-1.304	0.192	-0.360	0.072	C(job)[T.unemployed]	-0.1438	0.110	-1.304	0.192	-0.360	0.072
15	C(job)[T.unknown]	-0.2998	0.230	-1.302	0.193	-0.751	0.151	C(job)[T.unknown]	-0.2998	0.230	-1.302	0.193	-0.751	0.151
16	C(month)[T.aug]	-0.7064	0.078	-9.036	0.000	-0.860	-0.553	C(month)[T.aug]	-0.7064	0.078	-9.036	0.000	-0.860	-0.553
17	C(month)[T.dec]	0.7147	0.176	4.053	0.000	0.369	1.060	C(month)[T.dec]	0.7147	0.176	4.053	0.000	0.369	1.060
18	C(month)[T.feb]	-0.1413	0.089	-1.583	0.113	-0.316	0.034	C(month)[T.feb]	-0.1413	0.089	-1.583	0.113	-0.316	0.034
19	C(month)[T.jan]	-1.2649	0.121	-10.412	0.000	-1.503	-1.027	C(month)[T.jan]	-1.2649	0.121	-10.412	0.000	-1.503	-1.027
20	C(month)[T.jul]	-0.9189	0.077	-11.995	0.000	-1.069	-0.769	C(month)[T.jul]	-0.9189	0.077	-11.995	0.000	-1.069	-0.769
21	C(month)[T.jun]	0.4662	0.094	4.980	0.000	0.283	0.650	C(month)[T.jun]	0.4662	0.094	4.980	0.000	0.283	0.650
22	C(month)[T.mar]	1.6243	0.119	13.595	0.000	1.390	1.858	C(month)[T.mar]	1.6243	0.119	13.595	0.000	1.390	1.858
23	C(month)[T.may]	-0.3804	0.072	-5.281	0.000	-0.522	-0.239	C(month)[T.may]	-0.3804	0.072	-5.281	0.000	-0.522	-0.239
24	C(month)[T.nov]	-0.9173	0.084	-10.905	0.000	-1.082	-0.752	C(month)[T.nov]	-0.9173	0.084	-10.905	0.000	-1.082	-0.752
25	C(month)[T.oct]	0.8956	0.108	8.293	0.000	0.684	1.107	C(month)[T.oct]	0.8956	0.108	8.293	0.000	0.684	1.107
26	C(month)[T.sep]	0.8829	0.119	7.392	0.000	0.649	1.117	C(month)[T.sep]	0.8829	0.119	7.392	0.000	0.649	1.117
27	C(poutcome)[T.other]	0.3333	0.169	1.970	0.049	0.002	0.665	C(poutcome)[T.other]	0.3333	0.169	1.970	0.049	0.002	0.665
28	C(poutcome)[T.success]	2.4414	0.160	15.269	0.000	2.128	2.755	C(poutcome)[T.success]	2.4414	0.160	15.269	0.000	2.128	2.755
29	C(poutcome)[T.unknown]	-0.0414	0.227	-0.183	0.855	-0.486	0.403	C(poutcome)[T.unknown]	-0.0414	0.227	-0.183	0.855	-0.486	0.403
30	balance	1.525e-05	5.09e-06	2.996	0.003	5.27e-06	2.52e-05	balance	0.0464	0.015	2.996	0.003	0.016	0.077
31	day	0.0107	0.002	4.281	0.000	0.006	0.016	day	0.0889	0.021	4.281	0.000	0.048	0.130
32	duration	0.0042	6.42e-05	65.169	0.000	0.004	0.004	duration	1.0783	0.017	65.169	0.000	1.046	1.111
33	campaign	-0.0940	0.010	-9.206	0.000	-0.114	-0.074	campaign	-0.2912	0.032	-9.206	0.000	-0.353	-0.229
34	pdays	0.0002	0.000	0.413	0.679	-0.001	0.001	pdays	0.0190	0.046	0.413	0.679	-0.071	0.109
35	pdays:C(poutcome)[T.other]	-0.0003	0.001	-0.359	0.719	-0.002	0.001	pdays:C(poutcome)[T.other]	-0.0266	0.074	-0.359	0.719	-0.172	0.119
36	pdays:C(poutcome)[T.success]	-0.0004	0.001	-0.584	0.560	-0.002	0.001	pdays:C(poutcome)[T.success]	-0.0444	0.076	-0.584	0.560	-0.194	0.105
37	pdays:C(poutcome)[T.unknown]	0.0041	0.008	0.523	0.601	-0.011	0.020	pdays:C(poutcome)[T.unknown]	0.4130	0.790	0.523	0.601	-1.136	1.962
38	previous	0.0416	0.021	1.945	0.052	-0.000	0.084	previous	0.0959	0.049	1.945	0.052	-0.001	0.193
39	previous:C(poutcome)[T.other]	-0.0269	0.017	-1.610	0.107	-0.060	0.006	previous:C(poutcome)[T.other]	-0.0621	0.039	-1.610	0.107	-0.138	0.013
40	previous:C(poutcome)[T.success]	-0.0179	0.030	-0.606	0.544	-0.076	0.040	previous:C(poutcome)[T.success]	-0.0413	0.068	-0.606	0.544	-0.175	0.092
41	previous:C(poutcome)[T.unknown]	-0.4841	0.694	-0.697	0.486	-1.844	0.876	previous:C(poutcome)[T.unknown]	-1.1150	1.599	-0.697	0.486	-4.248	2.018
42	pdays:previous	-4.709e-05	7.42e-05	-0.634	0.526	-0.000	9.84e-05	pdays:previous	-0.0109	0.017	-0.634	0.526	-0.044	0.023

 

Variance inflation factors

features with multi-collinearity (vif > 10): default, marital, loan, education, age

feature	variance inflation factor without target	variance inflation factor with target	ranking
default	89.590922	90.757180	1.0
marital	34.094449	34.095701	2.0
loan	29.192732	29.204954	3.0
education	27.754247	27.754449	4.0
age	18.582507	18.598972	5.0
job	9.713535	9.744778	6.0
housing	9.418423	9.458427	7.0
contact	8.074904	8.092028	8.0
day	4.757858	4.758009	9.0
month	3.446856	3.559682	10.0
poutcome	2.899358	3.101371	11.0
duration	2.022798	2.419260	12.0
campaign	1.873813	1.874478	13.0
pdays	1.721085	1.721512	14.0
y	-	1.597047	15.0
previous	1.373109	1.373142	16.0
balance	1.228526	1.229092	17.0

 

 

 

 

Covariance Analysis

correlation of age, balance, day, duration campaign, pdays, previous

As shown in the heatmap, "deposit yes" exhibits strong correlations with customer attributes. I have performed regression analysis and principle component analysis(PCA) to explore impact of individual attributes on "deposit yes". 

 

 

Principle component analysis

• Features with high-variance : balance, age, day, duration, campaign, pdays, previous
• Selected strongly correlated features
  • Correlation between explainatory features 
    • positive correlation: (job, education), (pdays, previous), (pdays, poutcome), (previous, poutcome)
    • negative correlation: (age, marital)
  • Target feature correlation
    • positive correlation: duration, poutcome, month
• Efficient feature dimension range: 7 ~ 9

 

Regression coefficient without feature interaction

(Note) Covariance analysis has been conducted without one-hot encoding for categorical variables.  

	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15
centering	-0.0	4.0	4.0	3.0	0.0	0.0	6.0	4.0	7.0	-0.0	4.0	0.0	-0.0	0.0	0.0	4.0
standardizing	-0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0	-0.0	0.0	1.0	-0.0	0.0	0.0	0.0
pc_centering	0.0	0.0	0.0	0.0	-0.0	-0.0	0.0	-8.0	4.0	-9.0	-0.0	0.0	-4.0	-2.0	2.0	-0.0
pc_standardizing	1.0	-0.0	-0.0	-0.0	-0.0	1.0	-0.0	-0.0	-0.0	0.0	-0.0	-0.0	-0.0	-0.0	0.0	0.0

centering & standardizing : '0:age', '1:job', '2:marital', '3:education', '4:default', '5:balance', '6:housing', '7:loan', '8:contact', '9:day', '10:month', '11:duration', '12:campaign', '13:pdays', '14:previous', '15:poutcome'

 

 

Explainatory Centering Features with target feature

Principle Component of Explainatory Centering Features without target feature

Principle Component of Explainatory Standardizing Features without target feature

 

 

Exploratory factor analysis

: (orthogonal) varimax rotation

Factor Loadings with target feature

Factor Loadings without target feature

 

---

Exploratory Data Analysis

Exploration of association with target feature: multivariate analysis

Conditional Probability for deposit(y; target) feature, conditions = (duration, pcoutcome)

 

	y	Level0			duration	Level1			poutcome	Level2
		count	probaility	rank		count	probaility	rank		count	probaility	rank
0	no	39922	0.883015	1.0	(-4.918, 491.8]	36702	0.811794	1.0	failure	3990	0.088253	2.0
1	no	39922	0.883015	1.0	(-4.918, 491.8]	36702	0.811794	1.0	other	1407	0.031121	5.0
2	no	39922	0.883015	1.0	(-4.918, 491.8]	36702	0.811794	1.0	success	486	0.010750	8.0
3	no	39922	0.883015	1.0	(-4.918, 491.8]	36702	0.811794	1.0	unknown	30819	0.681670	1.0
4	no	39922	0.883015	1.0	(1475.4, 1967.2]	64	0.001416	8.0	failure	6	0.000133	31.5
5	no	39922	0.883015	1.0	(1475.4, 1967.2]	64	0.001416	8.0	other	3	0.000066	37.0
6	no	39922	0.883015	1.0	(1475.4, 1967.2]	64	0.001416	8.0	success	1	0.000022	42.0
7	no	39922	0.883015	1.0	(1475.4, 1967.2]	64	0.001416	8.0	unknown	54	0.001194	19.0
8	no	39922	0.883015	1.0	(1967.2, 2459.0]	20	0.000442	10.0	failure	2	0.000044	38.0
9	no	39922	0.883015	1.0	(1967.2, 2459.0]	20	0.000442	10.0	other	1	0.000022	42.0
10	no	39922	0.883015	1.0	(1967.2, 2459.0]	20	0.000442	10.0	unknown	17	0.000376	27.0
11	no	39922	0.883015	1.0	(2459.0, 2950.8]	4	0.000088	14.0	unknown	4	0.000088	35.5
12	no	39922	0.883015	1.0	(2950.8, 3442.6]	6	0.000133	12.0	unknown	6	0.000133	31.5
13	no	39922	0.883015	1.0	(3442.6, 3934.4]	1	0.000022	16.0	unknown	1	0.000022	42.0
14	no	39922	0.883015	1.0	(4426.2, 4918.0]	1	0.000022	16.0	unknown	1	0.000022	42.0
15	no	39922	0.883015	1.0	(491.8, 983.6]	2776	0.061401	3.0	failure	249	0.005508	12.0
16	no	39922	0.883015	1.0	(491.8, 983.6]	2776	0.061401	3.0	other	104	0.002300	16.0
17	no	39922	0.883015	1.0	(491.8, 983.6]	2776	0.061401	3.0	success	39	0.000863	20.0
18	no	39922	0.883015	1.0	(491.8, 983.6]	2776	0.061401	3.0	unknown	2384	0.052731	3.0
19	no	39922	0.883015	1.0	(983.6, 1475.4]	348	0.007697	6.0	failure	36	0.000796	21.0
20	no	39922	0.883015	1.0	(983.6, 1475.4]	348	0.007697	6.0	other	18	0.000398	26.0
21	no	39922	0.883015	1.0	(983.6, 1475.4]	348	0.007697	6.0	success	7	0.000155	29.5
22	no	39922	0.883015	1.0	(983.6, 1475.4]	348	0.007697	6.0	unknown	287	0.006348	11.0
23	yes	5289	0.116985	2.0	(-4.918, 491.8]	2975	0.065803	2.0	failure	406	0.008980	10.0
24	yes	5289	0.116985	2.0	(-4.918, 491.8]	2975	0.065803	2.0	other	208	0.004601	13.0
25	yes	5289	0.116985	2.0	(-4.918, 491.8]	2975	0.065803	2.0	success	782	0.017297	7.0
26	yes	5289	0.116985	2.0	(-4.918, 491.8]	2975	0.065803	2.0	unknown	1579	0.034925	4.0
27	yes	5289	0.116985	2.0	(1475.4, 1967.2]	112	0.002477	7.0	failure	9	0.000199	28.0
28	yes	5289	0.116985	2.0	(1475.4, 1967.2]	112	0.002477	7.0	other	4	0.000088	35.5
29	yes	5289	0.116985	2.0	(1475.4, 1967.2]	112	0.002477	7.0	success	5	0.000111	33.5
30	yes	5289	0.116985	2.0	(1475.4, 1967.2]	112	0.002477	7.0	unknown	94	0.002079	17.0
31	yes	5289	0.116985	2.0	(1967.2, 2459.0]	23	0.000509	9.0	failure	1	0.000022	42.0
32	yes	5289	0.116985	2.0	(1967.2, 2459.0]	23	0.000509	9.0	success	1	0.000022	42.0
33	yes	5289	0.116985	2.0	(1967.2, 2459.0]	23	0.000509	9.0	unknown	21	0.000464	24.0
34	yes	5289	0.116985	2.0	(2459.0, 2950.8]	7	0.000155	11.0	unknown	7	0.000155	29.5
35	yes	5289	0.116985	2.0	(2950.8, 3442.6]	5	0.000111	13.0	unknown	5	0.000111	33.5
36	yes	5289	0.116985	2.0	(3442.6, 3934.4]	1	0.000022	16.0	unknown	1	0.000022	42.0
37	yes	5289	0.116985	2.0	(491.8, 983.6]	1649	0.036473	4.0	failure	170	0.003760	14.0
38	yes	5289	0.116985	2.0	(491.8, 983.6]	1649	0.036473	4.0	other	76	0.001681	18.0
39	yes	5289	0.116985	2.0	(491.8, 983.6]	1649	0.036473	4.0	success	167	0.003694	15.0
40	yes	5289	0.116985	2.0	(491.8, 983.6]	1649	0.036473	4.0	unknown	1236	0.027338	6.0
41	yes	5289	0.116985	2.0	(983.6, 1475.4]	517	0.011435	5.0	failure	32	0.000708	22.0
42	yes	5289	0.116985	2.0	(983.6, 1475.4]	517	0.011435	5.0	other	19	0.000420	25.0
43	yes	5289	0.116985	2.0	(983.6, 1475.4]	517	0.011435	5.0	success	23	0.000509	23.0
44	yes	5289	0.116985	2.0	(983.6, 1475.4]	517	0.011435	5.0	unknown	443	0.009799	9.0

		support			confidence		lift
		CondFreq	no	yes	no	yes	no	yes
duration	poutcome
(-4.918, 491.8]	unknown	32398.0	30819.0	1579.0	0.951262	0.048738	1.077289	0.416615
	failure	4396.0	3990.0	406.0	0.907643	0.092357	1.027891	0.789476
	other	1615.0	1407.0	208.0	0.871207	0.128793	0.986628	1.100934
	success	1268.0	486.0	782.0	0.383281	0.616719	0.434059	5.271789
(491.8, 983.6]	unknown	3620.0	2384.0	1236.0	0.658564	0.341436	0.745812	2.918639
	failure	419.0	249.0	170.0	0.594272	0.405728	0.673003	3.468210
	other	180.0	104.0	76.0	0.577778	0.422222	0.654324	3.609206
	success	206.0	39.0	167.0	0.189320	0.810680	0.214402	6.929786
(1475.4, 1967.2]	unknown	148.0	54.0	94.0	0.364865	0.635135	0.413203	5.429211
	failure	15.0	6.0	9.0	0.400000	0.600000	0.452993	5.128871
	other	7.0	3.0	4.0	0.428571	0.571429	0.485350	4.884639
	success	6.0	1.0	5.0	0.166667	0.833333	0.188747	7.123432
(983.6, 1475.4]	unknown	730.0	287.0	443.0	0.393151	0.606849	0.445237	5.187420
	failure	68.0	36.0	32.0	0.529412	0.470588	0.599550	4.022644
	other	37.0	18.0	19.0	0.486486	0.513514	0.550938	4.389574
	success	30.0	7.0	23.0	0.233333	0.766667	0.264246	6.553558
(1967.2, 2459.0]	unknown	38.0	17.0	21.0	0.447368	0.552632	0.506637	4.723960
	failure	3.0	2.0	1.0	0.666667	0.333333	0.754989	2.849373
	other	1.0	1.0	0.0	1.000000	0.000000	1.132483	0.000000
	success	1.0	0.0	1.0	0.000000	1.000000	0.000000	8.548119
(2459.0, 2950.8]	unknown	11.0	4.0	7.0	0.363636	0.636364	0.411812	5.439712
	failure	0.0	0.0	0.0	0.000000	0.000000	NaN	NaN
	other	0.0	0.0	0.0	0.000000	0.000000	NaN	NaN
	success	0.0	0.0	0.0	0.000000	0.000000	NaN	NaN
(2950.8, 3442.6]	unknown	11.0	6.0	5.0	0.545455	0.454545	0.617718	3.885509
	failure	0.0	0.0	0.0	0.000000	0.000000	NaN	NaN
	other	0.0	0.0	0.0	0.000000	0.000000	NaN	NaN
	success	0.0	0.0	0.0	0.000000	0.000000	NaN	NaN
(3442.6, 3934.4]	unknown	2.0	1.0	1.0	0.500000	0.500000	0.566242	4.274059
	failure	0.0	0.0	0.0	0.000000	0.000000	NaN	NaN
	other	0.0	0.0	0.0	0.000000	0.000000	NaN	NaN
	success	0.0	0.0	0.0	0.000000	0.000000	NaN	NaN
(4426.2, 4918.0]	unknown	1.0	1.0	0.0	1.000000	0.000000	1.132483	0.000000
	failure	0.0	0.0	0.0	0.000000	0.000000	NaN	NaN
	other	0.0	0.0	0.0	0.000000	0.000000	NaN	NaN
	success	0.0	0.0	0.0	0.000000	0.000000	NaN	NaN

 

Feature importance and information values for target variable 

feature	feature importance	fi ranking	information values	iv ranking
duration	0.297104	1.0	1.816187	1
balance	0.120111	2.0	1.280381	2
poutcome	0.098661	3.0	0.514609	4
month	0.091646	4.0	0.436131	5
age	0.091354	5.0	0.226923	8
day	0.084840	6.0	0.117758	11
job	0.045463	7.0	0.155697	10
pdays	0.036583	8.0	0.559210	3
campaign	0.033835	9.0	0.089986	12
contact	0.023641	10.0	0.300396	6
education	0.020378	11.0	0.050112	14
marital	0.016275	12.0	0.040127	15
previous	0.016039	13.0	0.230969	7
housing	0.015632	14.0	0.188681	9
loan	0.007100	15.0	0.054859	13
default	0.001339	16.0	0.006256	16

Decision path about target variable

Decision Tree(criterion='gini', min_impurity_decrease=0.001); Decision Path

 

Exploration of independence between features: bivariate analysis

Decision boundary formation by non-linear transformation

The variables that has the most influence on the target variable: duration(numerical feature), poutcome(categorical feature) 

		chi2				mi
		statistic	p-value	dof	dependence	mi	adjusted_mi	normalized_mi
job	marital	3837.6027	0.0000	22	True	0.0437	0.0287	0.0289
	education	28483.1365	0.0000	33	True	0.3069	0.1894	0.1896
	default	60.3425	0.0000	11	True	0.0007	0.0005	0.0006
	housing	3588.7309	0.0000	11	True	0.0411	0.0292	0.0293
	loan	512.8105	0.0000	11	True	0.0069	0.0053	0.0054
	contact	2047.1332	0.0000	22	True	0.0207	0.0139	0.0141
	month	6043.8664	0.0000	121	True	0.0630	0.0297	0.0304
	poutcome	559.2778	0.0000	33	True	0.0058	0.0039	0.0042
	y	836.1055	0.0000	11	True	0.0083	0.0066	0.0067
marital	education	1337.5099	0.0000	6	True	0.0161	0.0158	0.0159
	default	16.7194	0.0002	2	True	0.0002	0.0003	0.0003
	housing	19.3448	0.0001	2	True	0.0002	0.0002	0.0003
	loan	121.9525	0.0000	2	True	0.0014	0.0020	0.0021
	contact	183.8431	0.0000	4	True	0.0021	0.0023	0.0024
	month	472.8791	0.0000	22	True	0.0052	0.0034	0.0035
	poutcome	76.4791	0.0000	6	True	0.0008	0.0010	0.0011
	y	196.4959	0.0000	2	True	0.0021	0.0033	0.0033
education	default	11.4246	0.0096	3	True	0.0001	0.0002	0.0002
	housing	643.8888	0.0000	3	True	0.0071	0.0078	0.0079
	loan	291.3714	0.0000	3	True	0.0035	0.0044	0.0044
	contact	1363.4366	0.0000	6	True	0.0151	0.0155	0.0156
	month	1644.2895	0.0000	33	True	0.0178	0.0110	0.0113
	poutcome	172.3951	0.0000	9	True	0.0019	0.0020	0.0022
	y	238.9235	0.0000	3	True	0.0026	0.0035	0.0035
default	housing	1.5514	0.2129	1	False	0.0000	0.0000	0.0000
	loan	268.1092	0.0000	1	True	0.0024	0.0089	0.0089
	contact	26.9295	0.0000	2	True	0.0003	0.0007	0.0007
	month	155.6489	0.0000	11	True	0.0021	0.0019	0.0020
	poutcome	73.8033	0.0000	3	True	0.0011	0.0029	0.0030
	y	22.2022	0.0000	1	True	0.0003	0.0013	0.0013
housing	loan	76.9748	0.0000	1	True	0.0009	0.0015	0.0015
	contact	2062.4619	0.0000	2	True	0.0235	0.0312	0.0312
	month	11494.0192	0.0000	11	True	0.1396	0.1024	0.1025
	poutcome	926.4237	0.0000	3	True	0.0105	0.0156	0.0157
	y	874.8224	0.0000	1	True	0.0097	0.0184	0.0184
loan	contact	11.9735	0.0025	2	True	0.0001	0.0002	0.0002
	month	1511.2025	0.0000	11	True	0.0155	0.0124	0.0125
	poutcome	137.9993	0.0000	3	True	0.0019	0.0035	0.0035
	y	209.6170	0.0000	1	True	0.0026	0.0065	0.0066
contact	month	23715.3268	0.0000	22	True	0.2971	0.2082	0.2083
	poutcome	3892.1528	0.0000	6	True	0.0625	0.0851	0.0852
	y	1035.7142	0.0000	2	True	0.0136	0.0231	0.0232
month	poutcome	6230.9857	0.0000	33	True	0.0638	0.0472	0.0475
	y	3061.8389	0.0000	11	True	0.0244	0.0202	0.0203
poutcome	y	4391.5066	0.0000	3	True	0.0294	0.0581	0.0582

		correlation							regression
		pearson	p-pval	spearmanr	s-pval	kendalltau	k-pval	quasi-dependence	f	f-pval	quasi-dependence
age	balance	0.097783	1.846987e-96	0.096380	9.361066e-94	0.065226	6.014181e-93	True	436.437210	1.846987e-96	True
	day	-0.009120	5.248053e-02	-0.008948	5.709532e-02	-0.006681	3.907395e-02	True	3.760582	5.248053e-02	False
	duration	-0.004648	3.229726e-01	-0.033257	1.514473e-12	-0.022444	1.784082e-12	True	0.976892	3.229726e-01	False
	campaign	0.004760	3.114630e-01	0.037136	2.816751e-15	0.027816	2.757255e-15	True	1.024485	3.114630e-01	False
	pdays	-0.023758	4.367248e-07	-0.017468	2.036697e-04	-0.013679	2.356496e-04	True	25.532326	4.367248e-07	True
	previous	0.001288	7.841413e-01	-0.011900	1.139584e-02	-0.009518	1.129966e-02	True	0.075037	7.841413e-01	False
balance	day	0.004503	3.383868e-01	0.001329	7.775064e-01	0.001242	6.982198e-01	False	0.916553	3.383868e-01	False
	duration	0.021560	4.545003e-06	0.042651	1.161677e-19	0.028586	1.086553e-19	True	21.025178	4.545003e-06	True
	campaign	-0.014578	1.936247e-03	-0.030959	4.573514e-11	-0.022924	4.563415e-11	True	9.610140	1.936247e-03	True
	pdays	0.003435	4.651272e-01	0.069676	9.007228e-50	0.054180	4.248024e-49	True	0.533537	4.651272e-01	False
	previous	0.016674	3.919530e-04	0.079536	2.361550e-64	0.062863	3.301276e-64	True	12.572057	3.919530e-04	True
day	duration	-0.030206	1.327167e-10	-0.058142	3.673297e-35	-0.039337	8.186136e-35	True	41.287405	1.327167e-10	True
	campaign	0.162490	4.793707e-265	0.139581	1.892587e-195	0.105353	3.056587e-195	True	1226.027295	4.793707e-265	True
	pdays	-0.093044	1.764882e-87	-0.092226	5.644265e-86	-0.072813	1.148434e-84	True	394.801213	1.764882e-87	True
	previous	-0.051710	3.729346e-28	-0.087780	4.944255e-78	-0.070418	8.956173e-78	True	121.211875	3.729346e-28	True
duration	campaign	-0.084570	1.521417e-72	-0.107962	2.779222e-117	-0.079976	3.223840e-117	True	325.663953	1.521417e-72	True
	pdays	-0.001565	7.393560e-01	0.028698	1.040577e-09	0.022478	9.221545e-10	True	0.110695	7.393560e-01	False
	previous	0.001203	7.981072e-01	0.031175	3.355401e-11	0.024689	2.785197e-11	True	0.065433	7.981072e-01	False
campaign	pdays	-0.088628	1.621197e-79	-0.112284	9.254439e-127	-0.096802	1.246977e-125	True	357.921953	1.621197e-79	True
	previous	-0.032855	2.794818e-12	-0.108448	2.496456e-118	-0.094371	3.617957e-117	True	48.854499	2.794818e-12	True
pdays	previous	0.454820	0.000000e+00	0.985645	0.000000e+00	0.902709	0.000000e+00	True	11791.089955	0.000000e+00	True

		variance			correlation
		f	f-pval	quasi_dependence	spearmanr	s-pval	kendalltau	k-pval	quasi_dependence
categorical	numerical
job	age	1377.936493	0.000000e+00	True	-0.008217	8.062405e-02	-0.003349	3.241846e-01	False
	balance	43.007783	5.709430e-94	True	0.029609	3.036066e-10	0.021057	3.651549e-10	True
	day	9.335477	5.489892e-17	True	0.022320	2.070497e-06	0.016542	1.230417e-06	True
	duration	6.842766	1.232447e-11	True	0.005277	2.618159e-01	0.003780	2.596166e-01	False
	campaign	12.483647	6.253473e-24	True	0.012609	7.337685e-03	0.009946	7.306402e-03	True
	pdays	14.161079	1.107741e-27	True	-0.008851	5.984944e-02	-0.007197	6.621746e-02	False
	previous	7.591359	3.183471e-13	True	-0.002165	6.452470e-01	-0.001853	6.396465e-01	False
marital	age	5228.732920	0.000000e+00	True	-0.442815	0.000000e+00	-0.354618	0.000000e+00	True
	balance	17.954318	1.605587e-08	True	0.020281	1.612420e-05	0.015796	2.068852e-05	True
	day	1.348193	2.597196e-01	False	-0.006203	1.871711e-01	-0.004938	1.898360e-01	False
	duration	12.078630	5.697950e-06	True	0.017361	2.229138e-04	0.013683	2.198743e-04	True
	campaign	22.336983	2.013545e-10	True	-0.030345	1.092802e-10	-0.026402	1.140811e-10	True
	pdays	19.695866	2.817855e-09	True	0.025644	4.942493e-08	0.023614	4.831779e-08	True
	previous	6.550023	1.431440e-03	True	0.025697	4.637298e-08	0.023874	4.695176e-08	True
education	age	731.757745	0.000000e+00	True	-0.115575	3.122264e-134	-0.090377	1.833152e-133	True
	balance	116.682074	2.849538e-75	True	0.075328	6.801877e-58	0.058231	9.724104e-58	True
	day	10.166018	1.089429e-06	True	0.024587	1.708703e-07	0.019347	1.587951e-07	True
	duration	0.218271	8.837767e-01	False	-0.003701	4.312879e-01	-0.002875	4.281401e-01	False
	campaign	6.617783	1.824042e-04	True	-0.001645	7.265724e-01	-0.001410	7.253331e-01	False
	pdays	8.746901	8.522341e-06	True	0.026293	2.252637e-08	0.023545	2.804749e-08	True
	previous	10.362132	8.192732e-07	True	0.034730	1.505560e-13	0.031632	1.510394e-13	True
default	age	14.456560	1.436177e-04	True	-0.014681	1.798204e-03	-0.012157	1.798915e-03	True
	balance	202.302934	8.246278e-46	True	-0.167739	1.495206e-282	-0.137371	1.345449e-278	True
	day	4.015362	4.509349e-02	True	0.009727	3.862282e-02	0.008087	3.862420e-02	True
	duration	4.540782	3.310185e-02	True	-0.007100	1.311333e-01	-0.005803	1.311318e-01	False
	campaign	12.796137	3.476985e-04	True	0.014265	2.419778e-03	0.012894	2.420612e-03	True
	pdays	40.668687	1.820913e-10	True	-0.038053	5.780955e-16	-0.036344	5.915284e-16	True
	previous	15.193840	9.715925e-05	True	-0.039279	6.554908e-17	-0.037892	6.728462e-17	True
housing	age	1611.326374	0.000000e+00	True	-0.154340	5.071808e-239	-0.127809	3.390580e-236	True
	balance	214.812902	1.582632e-48	True	-0.068292	7.020174e-48	-0.055928	8.962332e-48	True
	day	35.425150	2.669905e-09	True	-0.027605	4.340367e-09	-0.022951	4.367189e-09	True
	duration	1.164622	2.805147e-01	False	0.005187	2.700684e-01	0.004240	2.700637e-01	False
	campaign	25.190874	5.212410e-07	True	-0.037807	8.877289e-16	-0.034174	9.078130e-16	True
	pdays	708.053596	8.305619e-155	True	0.080977	1.201289e-66	0.077341	1.950718e-66	True
	previous	62.231686	3.121519e-15	True	0.062087	7.288802e-40	0.059896	8.608649e-40	True
loan	age	11.082880	8.719799e-04	True	-0.004720	3.155434e-01	-0.003909	3.155381e-01	False
	balance	323.965408	3.544641e-72	True	-0.128966	6.474347e-167	-0.105618	1.515890e-165	True
	day	5.845397	1.562178e-02	True	0.012205	9.455904e-03	0.010147	9.457379e-03	True
	duration	6.965838	8.310901e-03	True	-0.013211	4.967530e-03	-0.010798	4.968703e-03	True
	campaign	4.503144	3.383802e-02	True	0.001587	7.357456e-01	0.001435	7.357415e-01	False
	pdays	23.418093	1.307759e-06	True	-0.029571	3.197051e-10	-0.028243	3.223325e-10	True
	previous	5.514300	1.886590e-02	True	-0.030700	6.614993e-11	-0.029617	6.678465e-11	True
contact	age	677.227898	1.597830e-290	True	0.053128	1.249988e-29	0.042091	1.565797e-28	True
	balance	55.110597	1.244231e-24	True	-0.034245	3.256355e-13	-0.027321	3.537308e-13	True
	day	33.846140	2.050227e-15	True	-0.027426	5.457756e-09	-0.022262	5.305953e-09	True
	duration	19.925809	2.239459e-09	True	-0.036802	4.969902e-15	-0.029422	4.272230e-15	True
	campaign	70.326448	3.199084e-31	True	0.007996	8.909032e-02	0.007118	8.603679e-02	False
	pdays	1486.235447	0.000000e+00	True	-0.279500	0.000000e+00	-0.260481	0.000000e+00	True
	previous	550.425330	6.567538e-237	True	-0.278906	0.000000e+00	-0.262108	0.000000e+00	True
month	age	108.256036	3.059617e-245	True	-0.032608	4.062113e-12	-0.024028	2.017940e-12	True
	balance	102.277424	2.002290e-231	True	0.027575	4.512501e-09	0.018797	2.645153e-08	True
	day	1052.969050	0.000000e+00	True	0.006697	1.544308e-01	0.009692	4.715761e-03	True
	duration	19.028489	1.079679e-38	True	0.009111	5.270302e-02	0.006402	5.763659e-02	False
	campaign	232.959857	0.000000e+00	True	-0.147398	5.864119e-218	-0.116475	3.733046e-214	True
	pdays	365.011077	0.000000e+00	True	0.053558	4.388272e-30	0.043274	4.639729e-28	True
	previous	130.342323	3.718515e-296	True	0.056224	5.486129e-33	0.047424	9.785824e-33	True
poutcome	age	26.381925	4.840702e-17	True	0.013266	4.790407e-03	0.010502	5.682389e-03	True
	balance	23.570292	3.088104e-15	True	-0.075375	5.783112e-58	-0.060154	9.550789e-58	True
	day	113.814955	2.009226e-73	True	0.088062	1.591110e-78	0.071072	1.448078e-77	True
	duration	31.136681	4.250760e-20	True	-0.025125	9.140220e-08	-0.019909	1.085663e-07	True
	campaign	192.829765	2.888451e-124	True	0.116698	7.855655e-137	0.102844	6.674510e-136	True
	pdays	51189.981633	0.000000e+00	True	-0.990409	0.000000e+00	-0.933486	0.000000e+00	True
	previous	6179.512197	0.000000e+00	True	-0.987244	0.000000e+00	-0.925074	0.000000e+00	True
y	age	28.625233	8.825644e-08	True	-0.008750	6.281716e-02	-0.007246	6.281783e-02	False
	balance	126.572276	2.521114e-29	True	0.100295	2.095556e-101	0.082138	6.593767e-101	True
	day	36.359010	1.653880e-09	True	-0.029548	3.299041e-10	-0.024566	3.326067e-10	True
	duration	8333.761148	0.000000e+00	True	0.342469	0.000000e+00	0.279923	0.000000e+00	True
	campaign	243.358404	1.012347e-54	True	-0.084054	1.109367e-71	-0.075977	1.948470e-71	True
	pdays	490.696563	3.790553e-108	True	0.154055	3.900096e-238	0.147137	2.484050e-235	True
	previous	396.443989	7.801830e-88	True	0.169124	2.852229e-287	0.163155	3.491720e-283	True

 

Descriptive statistics

• Class imbalanced categorical features: default, loan ,y

Categorical Attributes

							count	ratio	rank	self-information
total_count	column	unique	top	freq	entropy	instance
45211	job	12	blue-collar	9732	3.055353	blue-collar	9732	0.215257	1.0	2.215866
						management	9458	0.209197	2.0	2.257067
						technician	7597	0.168034	3.0	2.573172
						admin.	5171	0.114375	4.0	3.128159
						services	4154	0.091880	5.0	3.444101
						retired	2264	0.050076	6.0	4.319728
						self-employed	1579	0.034925	7.0	4.839591
						entrepreneur	1487	0.032890	8.0	4.926197
						unemployed	1303	0.028820	9.0	5.116765
						housemaid	1240	0.027427	10.0	5.188262
						student	938	0.020747	11.0	5.590942
						unknown	288	0.006370	12.0	7.294461
	marital	3	married	27214	1.315270	married	27214	0.601933	1.0	0.732325
						single	12790	0.282896	2.0	1.821658
						divorced	5207	0.115171	3.0	3.118150
	education	4	secondary	23202	1.614902	secondary	23202	0.513194	1.0	0.962425
						tertiary	13301	0.294198	2.0	1.765139
						primary	6851	0.151534	3.0	2.722287
						unknown	1857	0.041074	4.0	4.605628
	default	2	no	44396	0.130212	no	44396	0.981973	1.0	0.026244
						yes	815	0.018027	2.0	5.793730
	housing	2	yes	25130	0.990985	yes	25130	0.555838	1.0	0.847263
						no	20081	0.444162	2.0	1.170843
	loan	2	no	37967	0.634851	no	37967	0.839774	1.0	0.251928
						yes	7244	0.160226	2.0	2.641815
	contact	3	cellular	29285	1.177525	cellular	29285	0.647741	1.0	0.626512
						unknown	13020	0.287983	2.0	1.795944
						telephone	2906	0.064276	3.0	3.959567
	month	12	may	13766	2.937381	may	13766	0.304483	1.0	1.715564
						jul	6895	0.152507	2.0	2.713051
						aug	6247	0.138174	3.0	2.855438
						jun	5341	0.118135	4.0	3.081492
						nov	3970	0.087810	5.0	3.509463
						apr	2932	0.064851	6.0	3.946717
						feb	2649	0.058592	7.0	4.093154
						jan	1403	0.031032	8.0	5.010087
						oct	738	0.016323	9.0	5.936909
						sep	579	0.012807	10.0	6.286967
						mar	477	0.010551	11.0	6.566541
						dec	214	0.004733	12.0	7.722919
	poutcome	4	unknown	36959	0.937015	unknown	36959	0.817478	1.0	0.290748
						failure	4901	0.108403	2.0	3.205526
						other	1840	0.040698	3.0	4.618896
						success	1511	0.033421	4.0	4.903098
	y	2	no	39922	0.520631	no	39922	0.883015	1.0	0.179490
						yes	5289	0.116985	2.0	3.095607

Numerical Attributes

	column	count	norm_statstic	norm_pval	normality	l_shift	r_shift	iqr_min	iqr_25	mean	iqr_75	iqr_max	std	diff_maxmin
0	age	45211.0	3066.989468	0.0	False	True	False	18.0	33.0	40.936210	48.0	95.0	10.618762	77.0
1	balance	45211.0	64697.210210	0.0	False	True	False	-8019.0	72.0	1362.272058	1428.0	102127.0	3044.765829	110146.0
2	day	45211.0	14624.380064	0.0	False	False	True	1.0	8.0	15.806419	21.0	31.0	8.322476	30.0
3	campaign	45211.0	45156.283654	0.0	False	True	False	1.0	1.0	2.763841	3.0	63.0	3.098021	62.0
4	pdays	45211.0	24050.969837	0.0	False	True	False	-1.0	-1.0	40.197828	-1.0	871.0	100.128746	872.0
5	previous	45211.0	134066.595245	0.0	False	True	False	0.0	0.0	0.580323	0.0	275.0	2.303441	275.0

 

Description for attributes 

1. age
2. job : type of job
3. marital : marital status
4. education
5. default: has credit in default?
6. housing: has housing loan?
7. loan: has personal loan? 
8. contact: contact communication type
9. month: last contact month of year
10. day_of_week: last contact day of the week
11. duration: last contact duration, in seconds (numeric). Important note: this attribute highly affects the output target (e.g., if duration=0 then y='no'). Yet, the duration is not known before a call is performed. Also, after the end of the call y is obviously known. Thus, this input should only be included for benchmark purposes and should be discarded if the intention is to have a realistic predictive model.
12. campaign: number of contacts performed during this campaign and for this client (numeric, includes last contact)
13. pdays: number of days that passed by after the client was last contacted from a previous campaign (numeric; 999 means client was not previously contacted)
14. previous: number of contacts performed before this campaign and for this client (numeric)
15. poutcome: outcome of the previous marketing campaign (categorical: 'failure','nonexistent','success') # social and economic context attributes
16. y - has the client subscribed a term deposit? (binary target: 'yes','no')

 

 

---

Data Extraction

bank-full.csv

4.40MB

bank.csv

0.44MB

	age	job	marital	education	default	balance	housing	loan	contact	day	month	duration	campaign	pdays	previous	poutcome	y
0	58	management	married	tertiary	no	2143	yes	no	unknown	5	may	261	1	-1	0	unknown	no
1	44	technician	single	secondary	no	29	yes	no	unknown	5	may	151	1	-1	0	unknown	no
2	33	entrepreneur	married	secondary	no	2	yes	yes	unknown	5	may	76	1	-1	0	unknown	no
3	47	blue-collar	married	unknown	no	1506	yes	no	unknown	5	may	92	1	-1	0	unknown	no
4	33	unknown	single	unknown	no	1	no	no	unknown	5	may	198	1	-1	0	unknown	no
5	35	management	married	tertiary	no	231	yes	no	unknown	5	may	139	1	-1	0	unknown	no
6	28	management	single	tertiary	no	447	yes	yes	unknown	5	may	217	1	-1	0	unknown	no
7	42	entrepreneur	divorced	tertiary	yes	2	yes	no	unknown	5	may	380	1	-1	0	unknown	no
8	58	retired	married	primary	no	121	yes	no	unknown	5	may	50	1	-1	0	unknown	no
9	43	technician	single	secondary	no	593	yes	no	unknown	5	may	55	1	-1	0	unknown	no

 

 

Missing value & Duplication inspection

	column		total	missing-value				duplication
		quasi-dtypes	freq	not_freq	freq	ratio	rank	cardinality	selectivity	rank
0	age	numeric	45211	45211	0	0.0	9.0	77	0.001703	14.0
1	job	string	45211	45211	0	0.0	9.0	12	0.000265	9.5
2	marital	string	45211	45211	0	0.0	9.0	3	0.000066	5.5
3	education	string	45211	45211	0	0.0	9.0	4	0.000088	7.5
4	default	string	45211	45211	0	0.0	9.0	2	0.000044	2.5
5	balance	numeric	45211	45211	0	0.0	9.0	7168	0.158545	17.0
6	housing	string	45211	45211	0	0.0	9.0	2	0.000044	2.5
7	loan	string	45211	45211	0	0.0	9.0	2	0.000044	2.5
8	contact	string	45211	45211	0	0.0	9.0	3	0.000066	5.5
9	day	numeric	45211	45211	0	0.0	9.0	31	0.000686	11.0
10	month	string	45211	45211	0	0.0	9.0	12	0.000265	9.5
11	duration	numeric	45211	45211	0	0.0	9.0	1573	0.034792	16.0
12	campaign	numeric	45211	45211	0	0.0	9.0	48	0.001062	13.0
13	pdays	numeric	45211	45211	0	0.0	9.0	559	0.012364	15.0
14	previous	numeric	45211	45211	0	0.0	9.0	41	0.000907	12.0
15	poutcome	string	45211	45211	0	0.0	9.0	4	0.000088	7.5
16	y	string	45211	45211	0	0.0	9.0	2	0.000044	2.5

---

References

• https://archive.ics.uci.edu/ml/datasets/Bank+Marketing