[NOTE] Algorithms in E-Commerce Company (Stitch Fix), talk by YuanBo

Stitch Fix

A DS (Data-Science) driven online shopping model. Customers receives boxes and choose to keep or return the items inside.

Culture

这个在电商模式上有所创新的美国电商，给予了数据科学非常重要的位置。从组织架构上DS部门的定位不是为其他部门服务，而是提供business insight和innovation. 另外，团队并未像通常一样按照技术智能划分为算法、数据等等，而是按照业务场景或者研究方向来组成纵向团队，以避免团队间的合作成本，例如往往当算法人员需要数据支持时，数据人员正在忙别的project。

Organisation: DS department is not a service org for other department but a team providing insight and new ideas.

Roles: people are not divided to groups by tech functions but to vertical teams focus on some specific topic. This avoids time wasted on team interaction and communication.

Process: not pure top-down like others companies but using a bi-directional process: DS provides insights and innovations while receiving top down orders.

Algo overview

Check this :
https://algorithms-tour.stitchfix.com/

It’s amazing

Talk of YuanBo

Check this: https://www.bilibili.com/video/BV1eV411r7AP

Intelligent warehousing

主要谈到只能仓储中，AGV的路径规划以及货架的最优摆放问题。

Order fullfillment using shadow price info

Key：如何在进行履约路由分配时考虑到有限库存带来的缺货风险

Check this paper:
Making Better Fulfillment Decisions on the Fly in an Online Retail Environment

Forecast

Boosting 决策树仍然是霸主般的存在。

Censored demand for forecasting: 矩阵回填. 将通过矩阵分解和回填来处理历史demand数据，然后再用以进行预测，有明显的效果提升。

背后思想：数据生成：bootstrapping, data augmentation, GAN

Advices

Data Scientist:

• Type1: data insight, pb definition, communication
• Type2: engineering, model iteration

Try to cover both profiles.