5.1. Bank Telemarketing Upsell Campaign

5.1.1. Summary

This dataset is related to a direct marketing campaigns (phone calls) of a Portuguese banking institution. The marketing campaigns were based on phone calls. Often, more than one contact to the same client was required, in order to access if the product (bank term deposit) would be or not subscribed.

Features recorded before the contact event are removed from the original dataset [1] to avoid data leakage. Only clients with a positive balance are considered, since clients in debt are not eligible for term deposits.

Classes	2
Subscribers	4787
Non-subscribers	33144
Samples	37931
Features	10

Other relevant information can be found in [2] and [3].

5.1.2. Using the Dataset

The dataset can be loaded through the load_upsell_bank_telemarketing function. This returns a Dataset object with the following attributes:

• data: the feature matrix

• target: the target vector

• tp_cost: the cost of a true positive

• fp_cost: the cost of a false positive

• fn_cost: the cost of a false negative

• tn_cost: the cost of a true negative

• feature_names: the feature names

• target_names: the target names

• DESCR: the full description of the dataset

from empulse.datasets import load_upsell_bank_telemarketing

dataset = load_upsell_bank_telemarketing()

Alternatively, the load function can also return the features, target, and costs separately, by setting return_X_y_costs=True. Additionally, you can specify that you want the output in a pandas.DataFrame format, by setting as_frame=True.

The following code snippet demonstrates how to load the dataset and fit a model using the CSLogitClassifier:

from empulse.datasets import load_upsell_bank_telemarketing
from empulse.models import CSLogitClassifier
from sklearn.compose import ColumnTransformer
from sklearn.pipeline import Pipeline
from sklearn.preprocessing import StandardScaler, TargetEncoder

X, y, tp_cost, fp_cost, fn_cost, tn_cost = load_upsell_bank_telemarketing(
    return_X_y_costs=True,
    as_frame=True
)
pipeline = Pipeline([
    ('preprocessor', ColumnTransformer([
        ('num', StandardScaler(), X.select_dtypes(include=['number']).columns),
        ('cat', TargetEncoder(), X.select_dtypes(include=['category']).columns)
    ])),
    ('model', CSLogitClassifier())
])
pipeline.fit(
    X,
    y,
    model__tp_cost=tp_cost,
    model__fp_cost=fp_cost,
    model__fn_cost=fn_cost,
    model__tn_cost=tn_cost
)

5.1.3. Cost Matrix

	Actual positive \(y_i = 1\)	Actual negative \(y_i = 0\)
Predicted positive \(\hat{y}_i = 1\)	tp_cost \(= c\)	fp_cost \(= c\)
Predicted negative \(\hat{y}_i = 0\)	fn_cost \(= r \cdot d_i \cdot b_i\)	tn_cost \(= 0\)

with

• \(c\) : cost of contacting the client

• \(r\) : interest rate of the term deposit

• \(d_i\) : fraction of the client’s balance that is deposited in the term deposit

• \(b_i\) : client’s balance

Using default parameters, it is assumed that \(c = 1\), \(r = 0.02463333\), \(d_i = 0.25\) for all clients. The default parameters are based on [4].

These assumptions can be changed by passing your own values to the load_upsell_bank_telemarketing function:

from empulse.datasets import load_upsell_bank_telemarketing

X, y, tp_cost, fp_cost, fn_cost, tn_cost = load_upsell_bank_telemarketing(
    return_X_y_costs=True,
    interest_rate=0.05,
    term_deposit_fraction=0.30,
    contact_cost=10,
)

5.1.4. Data Description

Variable Name	Description	Type
age	Age of the client	numeric
balance	Average yearly balance	numeric
previous	Number of contacts performed before this campaign and for this client	numeric
job	Type of job (e.g., ‘admin.’, ‘blue-collar’, ‘entrepreneur’, etc.)	categorical
marital	Marital status (‘divorced’, ‘married’, ‘single’)	categorical
education	Education level (‘primary’, ‘secondary’, ‘tertiary’, ‘unknown’)	categorical
has_credit_in_default	Has credit in default? (‘yes’ = 1, ‘no’ = 0)	binary
has_housing_loan	Has housing loan? (‘yes’ = 1, ‘no’ = 0)	binary
has_personal_loan	Has personal loan? (‘yes’ = 1, ‘no’ = 0)	binary
previous_outcome	Outcome of the previous marketing campaign (‘success’, ‘failure’, ‘other’, ‘unknown’)	categorical
subscribed	Has the client subscribed a term deposit? (‘yes’ = 1, ‘no’ = 0)	binary

5.1.5. References

[1]

Moro, S., Rita, P., & Cortez, P. (2014). Bank Marketing [Dataset]. UCI Machine Learning Repository. https://doi.org/10.24432/C5K306.

[2]

Moro, S., Cortez, P., & Rita, P. (2014). A data-driven approach to predict the success of bank telemarketing. Decision Support Systems, 62, 22-31.

[3]

S. Moro, R. Laureano and P. Cortez. Using Data Mining for Bank Direct Marketing: An Application of the CRISP-DM Methodology. In P. Novais et al. (Eds.), Proceedings of the European Simulation and Modelling Conference - ESM’2011, pp. 117-121, Guimaraes, Portugal, October, 2011. EUROSIS. [bank.zip]

[4]

A. Correa Bahnsen, A. Stojanovic, D.Aouada, B, Ottersten, “Improving Credit Card Fraud Detection with Calibrated Probabilities”, in Proceedings of the fourteenth SIAM International Conference on Data Mining, 677-685, 2014.