下载PDF
C5i > 实例探究 > Developed model to identify utility customers at risk of seasonal defection from Time-of-Use (TOU) price plan
C5i Logo

Developed model to identify utility customers at risk of seasonal defection from Time-of-Use (TOU) price plan

技术
  • 分析与建模 - 预测分析
适用行业
  • 公用事业
适用功能
  • 商业运营
用例
  • 补货预测
服务
  • 数据科学服务
挑战
The client, a large US public utility, was facing a challenge in identifying which customers on a Time-of-Use (TOU) price plan would switch to an alternate price plan. They needed to understand the factors that predict this switching behavior and create propensity scores for each of the current TOU price plan customers to identify their individual likelihood of switching to a different price plan. The goal was to correctly predict a likely switcher approximately 70% of the time.
关于客户
The customer is a large public utility company based in the United States. They provide essential utility services to a vast number of residential customers. The company offers various price plans to its customers, including a Time-of-Use (TOU) price plan. However, they were facing a challenge in identifying which customers on the TOU price plan were likely to switch to an alternate price plan. The company needed a solution that could accurately predict this switching behavior and help them retain their customers.
解决方案
Blueocean Market Intelligence developed a predictive analytics model using the client’s data repository of residential customer information and the Meter Data Management System over a two-year period. Historical and customer demographic information was used to predict program participation and identify differences between program switchers and non-switchers. Simultaneous models were run using C5.0, CHAID, QUEST, and logistic regression. The final commissioned model had a high level of predictive accuracy of 70% for who was most likely to switch.
运营影响
  • Developed a model to predict which customers currently on a Time-of-Use (TOU) price plan will switch to an alternate price plan
  • Identified key factors driving the switching behavior
  • Established profiles comparing Switchers to Non-Switchers on characteristics such as energy usage, savings and monthly bill amount
数量效益
  • The final commissioned model had a high level of predictive accuracy of 70% for who was most likely to switch

相关案例.

联系我们

欢迎与我们交流!

* Required
* Required
* Required
* Invalid email address
提交此表单,即表示您同意 IoT ONE 可以与您联系并分享洞察和营销信息。
不,谢谢,我不想收到来自 IoT ONE 的任何营销电子邮件。
提交

Thank you for your message!
We will contact you soon.