Decision Tree vs Job Graph: Combining Data Science and JTBD in One Project
An analyst from an insurance company reflects on how the same task of understanding customer needs is approached differently by product managers and data scientists. The author compares Decision Tree (ML) and Job Graph (JTBD) and uses an example of corporate health insurance to illustrate how a single purchase involves multiple stakeholders with different jobs, each with their own graph. The article explains the key differences between the two models and shows how examining a product from the job graph perspective reveals unmet needs.
The author, heading an analytics department in an insurance company, noticed a similarity between the 'job graph' concept from Advanced JTBD and the Decision Tree model from machine learning, but with opposite purposes. Decision Tree is a predictive model that classifies objects into one category, while Job Graph is an explanatory framework that reveals multiple motivations behind a single action. In a retail health insurance example, the client's simple action of buying a policy covers five different Big Jobs, but the company optimizes only one branch, failing others. For B2B corporate health insurance, there are at least three stakeholders—the company, the decision-maker (HR director), and the employee—each with their own job graph; a successful product must address all three. The author suggests that product analytics should measure which jobs are being fulfilled for which stakeholder, and that attrition analysis should consider all three graphs. They also highlight telemedicine as an example of a feature that accidentally covers additional jobs. The article concludes that Decision Tree is suitable for classification and prediction tasks, while Job Graph is useful for understanding motivations and product strategy.
- Abbreviations
- JTBD = Jobs To Be Done — Jobs To Be Done (работа, которую нужно сделать)
- ML = Machine Learning — машинное обучение
- ДМС = Voluntary Health Insurance — добровольное медицинское страхование
- B2B = Business to Business — бизнес для бизнеса
- ЛПР = Decision Maker — лицо, принимающее решения
- KPI = Key Performance Indicators — ключевые показатели эффективности
- HR = Human Resources — управление персоналом
- ФОТ = Payroll Fund — фонд оплаты труда
Source: Хабр — Data Mining —
original