One of the world’s largest AI OEM and a leading manufacturer of hi-tech electronic devices has a vision of improving their supply chain using a series of AI/ML applications. In an industry that was entrenched in mature contract manufacturing practices, this relatively young manufacturer partnered with Infosys to implement a product returns forecasting engine that significantly improves warranty provisioning, spare parts stocking, and service centre workforce planning.
- The objectives of the client were to:
- Implement an end-to-end ML based forecasting engine that predicts product returns during its warranty period
- Enable better warranty provisioning, optimize parts inventory and service centre staffing
Key Challenges
- Existing forecasting model based on Statistical techniques and involved manual effort
- Parametric model of forecasting- zero flexibility
- Need of a dataset agnostic solution that incorporates difference in sparsity, seasonality, geographical differences, return channels etc.
Ready to experience?
TALK TO EXPERTS
The Solution
Automated Predictive engine
- Ensemble based predictive engine that incorporates 5 different Predictive algorithms- Machine Learning, Weibull Distribution, Polynomial Regression, Non-parametric, Time-Series models
- Infosys developed an end-to-end completely automated pipeline that can :
- Designed an automated ML pipeline and a champion challenger framework that selects best of available 5 models, this helps obtain forecast from best model
All this in one single click.
Benefits
Automation enabled forecasts to be applied to different products without manual development, increasing SME productivity by 20x
Revision of inventory policies due to more accurate forecasts will result in reduced inventory carrying costs, estimated to the tune of $16 MUSD per annum
Scientific forecasts enabled warranty provisioning which was done based on average return rates earlier
