Bank Marketing Data Dictionary

Source

This dataset originates from a study by Portuguese researchers aiming to predict the outcome of calls that took place as part of a bank marketing campaign from 2008 to 2014:

A data-driven approach to predict the success of bank telemarketing By Sérgio Moro, P. Cortez, P. Rita. 2014

This paper was published in the journal Decision Support Systems

Several versions of this dataset exist. Moro 2014 uses a private dataset with much richer features for their analysis. Two simpler versions have been published on the UCI Machine Learning Repository, where they serve as benchmarks in the machine learning community. For this assignment, we are using the dataset in bank_marketing_dataset.csv.

This dataset combines some of the features found in the ‘bank-full.csv’ and ‘bank-additional-full.csv’ files on the UCI repo. In particular, it has all of the macroeconomic indicators present in ‘bank-additional-full’, the ‘year’, and the rest of the features are from ‘bank-full.csv’.

Variables

Variable	Role	Type	Description
age	Feature	Integer	Client age in years
job	Feature	Categorical	Occupation type (admin, blue-collar, entrepreneur, housemaid, management, retired, self-employed, services, student, technician, unemployed, unknown)
marital	Feature	Categorical	Marital status (married, divorced, single). Divorced includes widowed
education	Feature	Categorical	Education level (primary, secondary, tertiary, unknown)
default	Feature	Binary	Has credit in default?
balance	Feature	Integer	Average yearly balance in euros
housing	Feature	Binary	Has housing loan?
loan	Feature	Binary	Has personal loan?
contact	Feature	Categorical	Contact communication type (cellular, telephone, unknown)
month	Feature	Categorical	Last contact month of year
day	Feature	Integer	Day of call
duration	Feature	Integer	Duration of marketing call
campaign	Feature	Integer	Number of contacts during this campaign for this client
pdays	Feature	Integer	Days since last contact from a previous campaign. -1 means not previously contacted
previous	Feature	Integer	Number of contacts before this campaign for this client
poutcome	Feature	Categorical	Outcome of previous marketing campaign (unknown, failure, other, success)
month_numeric	Feature	Integer	Month of the call
year	Feature	Integer	Year of the call
euribor3m	Feature	Numeric	Euribor 3-month rate — daily indicator
emp.var.rate	Feature	Numeric	Employment variation rate in Portugal, change in unemployment rate from one quarter to the next — quarterly indicator
nr.employed	Feature	Numeric	Number of employees in Portugal — quarterly indicator (thousands)
cons.price.idx	Feature	Numeric	Consumer price index in Portugal — monthly indicator
cons.conf.idx	Feature	Numeric	Consumer confidence index in Portugal — monthly indicator
y	Target	Binary	Has the client subscribed a term deposit?