下载PDF
Kyvos Insights > 实例探究 > Material Forecasting on 650x More Data at a Global Sports Brand
Kyvos Insights Logo

Material Forecasting on 650x More Data at a Global Sports Brand

技术
  • 分析与建模 - 大数据分析
  • 基础设施即服务 (IaaS) - 云计算
  • 基础设施即服务 (IaaS) - 云存储服务
适用行业
  • 服装
  • 零售
适用功能
  • 采购
  • 产品研发
用例
  • 需求计划与预测
  • 库存管理
服务
  • 云规划/设计/实施服务
  • 数据科学服务
挑战
The leading apparel and footwear brand faced challenges in fine-tuning its material forecasting based on consumer demand patterns. With an extensive network of factories servicing global stores, it was difficult to estimate the exact quantity and type of raw materials for different manufacturing locations. The existing BI architecture did not allow them to analyze more than 18 weeks of data. They were pulling source data from Amazon S3 to Snowflake, building aggregates on Azure Analysis Services (AAS), and then performing analysis on Excel. This led to multiple points of failure and each hop had an associated cost. They were hitting the limits of AAS in terms of processing that could be done and missed SLAs due to high data volumes during the holiday season. As data volumes rose, Excel reports would often freeze/crash.
关于客户
The customer is a leading apparel and footwear brand with an extensive network of factories servicing global stores. They have a wide range of collections planned around seasons or times of the year. The brand wanted to fine-tune its material forecasting based on consumer demand patterns. They wanted to understand past patterns to project future demands. Granular details such as the amount sold by color, size, or style across countries or stores could further help improve the forecasting accuracy. However, their existing BI architecture did not allow them to analyze more than 18 weeks of data, which was a significant challenge.
解决方案
The brand wanted to migrate from AAS to eliminate the data volume limitations. They were looking for a solution that could work directly on S3 while delivering the same functionality as AAS. Kyvos helped them eliminate the inefficiencies in their BI architecture by building an OLAP layer directly on AWS. Advanced algorithms and cloud-native architecture helped build an OLAP cube on two years of data in an hour. They could deal with high-cardinality dimensions such as style or material and build aggregates as needed by the business. With Kyvos, there was no restriction on the amount of data that could be analyzed. They could plug in Excel directly into Kyvos and perform interactive analysis on the entire data without any latency. Since all aggregations were stored in the Kyvos layer, Excel queries became lightweight, and the reports and dashboards refreshed instantly.
运营影响
  • Eliminated multiple points of failure
  • Reduced costs associated with multiple hops in their BI architecture
  • High scalability provided the ability to cater to future data growth
数量效益
  • Ability to scale analytics from 18 weeks to 104 weeks
  • Material forecasting to the lowest level of detail
  • Ability to predict color-size level details

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.