Late payment prediction models for fair allocation of customer contact lists to call center agents

Jongmyoung Kim, Pilsung Kang

Research output: Contribution to journalArticle

8 Citations (Scopus)

Abstract

Debt collection via call centers is an important operation in many business domains since it can significantly improve a firm's financial status by turning bad receivables into normal cash income that contributes to profits. Since the job performance of call center agents who carry out debt collection is primarily evaluated by the amount of debt collected, call center managers are faced with the challenge of allocating customer contact lists in a fair manner to eliminate a non-controllable external factor that could distort the objective evaluation of the agent's job performance. In this paper, we develop five machine learning-based late payment prediction models and ten customer scoring rules to predict the payment likelihood and the amount of the late payment for the customers who currently have an unpaid debt. The proposed scoring rules are verified under ten different contexts by varying the number of agents. Experimental results confirm that the prediction model-based scoring rules lead to fairer customer allocation results among the agents compared to the existing heuristic-based customer scoring rules. Among the prediction models, a hybrid approach can capture the late payers effectively, whereas tree-based models report more impartial customer allocation than the other methods.

Original languageEnglish
Pages (from-to)84-101
Number of pages18
JournalDecision Support Systems
Volume85
DOIs
Publication statusPublished - 2016 May 1

Fingerprint

Learning systems
Profitability
Managers
Prediction model
Call centres
Fair allocation
Customer contact
Payment
Prediction
Call Centers
Debt
Scoring rules
Industry
Scoring
Work Performance
Job performance
Machine Learning
Heuristics
Cash
External factors

Keywords

  • Artificial neural network
  • Decision tree
  • Hybrid approach
  • Late payment prediction
  • Machine learning
  • Support vector machine

ASJC Scopus subject areas

  • Management Information Systems
  • Information Systems
  • Developmental and Educational Psychology
  • Arts and Humanities (miscellaneous)
  • Information Systems and Management

Cite this

Late payment prediction models for fair allocation of customer contact lists to call center agents. / Kim, Jongmyoung; Kang, Pilsung.

In: Decision Support Systems, Vol. 85, 01.05.2016, p. 84-101.

Research output: Contribution to journalArticle

@article{99753d64a3ae4a36a60112bfdb81c5e5,
title = "Late payment prediction models for fair allocation of customer contact lists to call center agents",
abstract = "Debt collection via call centers is an important operation in many business domains since it can significantly improve a firm's financial status by turning bad receivables into normal cash income that contributes to profits. Since the job performance of call center agents who carry out debt collection is primarily evaluated by the amount of debt collected, call center managers are faced with the challenge of allocating customer contact lists in a fair manner to eliminate a non-controllable external factor that could distort the objective evaluation of the agent's job performance. In this paper, we develop five machine learning-based late payment prediction models and ten customer scoring rules to predict the payment likelihood and the amount of the late payment for the customers who currently have an unpaid debt. The proposed scoring rules are verified under ten different contexts by varying the number of agents. Experimental results confirm that the prediction model-based scoring rules lead to fairer customer allocation results among the agents compared to the existing heuristic-based customer scoring rules. Among the prediction models, a hybrid approach can capture the late payers effectively, whereas tree-based models report more impartial customer allocation than the other methods.",
keywords = "Artificial neural network, Decision tree, Hybrid approach, Late payment prediction, Machine learning, Support vector machine",
author = "Jongmyoung Kim and Pilsung Kang",
year = "2016",
month = "5",
day = "1",
doi = "10.1016/j.dss.2016.03.002",
language = "English",
volume = "85",
pages = "84--101",
journal = "Decision Support Systems",
issn = "0167-9236",
publisher = "Elsevier",

}

TY - JOUR

T1 - Late payment prediction models for fair allocation of customer contact lists to call center agents

AU - Kim, Jongmyoung

AU - Kang, Pilsung

PY - 2016/5/1

Y1 - 2016/5/1

N2 - Debt collection via call centers is an important operation in many business domains since it can significantly improve a firm's financial status by turning bad receivables into normal cash income that contributes to profits. Since the job performance of call center agents who carry out debt collection is primarily evaluated by the amount of debt collected, call center managers are faced with the challenge of allocating customer contact lists in a fair manner to eliminate a non-controllable external factor that could distort the objective evaluation of the agent's job performance. In this paper, we develop five machine learning-based late payment prediction models and ten customer scoring rules to predict the payment likelihood and the amount of the late payment for the customers who currently have an unpaid debt. The proposed scoring rules are verified under ten different contexts by varying the number of agents. Experimental results confirm that the prediction model-based scoring rules lead to fairer customer allocation results among the agents compared to the existing heuristic-based customer scoring rules. Among the prediction models, a hybrid approach can capture the late payers effectively, whereas tree-based models report more impartial customer allocation than the other methods.

AB - Debt collection via call centers is an important operation in many business domains since it can significantly improve a firm's financial status by turning bad receivables into normal cash income that contributes to profits. Since the job performance of call center agents who carry out debt collection is primarily evaluated by the amount of debt collected, call center managers are faced with the challenge of allocating customer contact lists in a fair manner to eliminate a non-controllable external factor that could distort the objective evaluation of the agent's job performance. In this paper, we develop five machine learning-based late payment prediction models and ten customer scoring rules to predict the payment likelihood and the amount of the late payment for the customers who currently have an unpaid debt. The proposed scoring rules are verified under ten different contexts by varying the number of agents. Experimental results confirm that the prediction model-based scoring rules lead to fairer customer allocation results among the agents compared to the existing heuristic-based customer scoring rules. Among the prediction models, a hybrid approach can capture the late payers effectively, whereas tree-based models report more impartial customer allocation than the other methods.

KW - Artificial neural network

KW - Decision tree

KW - Hybrid approach

KW - Late payment prediction

KW - Machine learning

KW - Support vector machine

UR - http://www.scopus.com/inward/record.url?scp=84980052299&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84980052299&partnerID=8YFLogxK

U2 - 10.1016/j.dss.2016.03.002

DO - 10.1016/j.dss.2016.03.002

M3 - Article

AN - SCOPUS:84980052299

VL - 85

SP - 84

EP - 101

JO - Decision Support Systems

JF - Decision Support Systems

SN - 0167-9236

ER -