Skip to content
Artificial IntelligenceLiveDemand forecasting

Demand forecasting — Retail

Probabilistic forecasts at SKU and store level, feeding replenishment. Delivered for a retail client in Kolkata.

Location
Kolkata
Year
2020
Duration
29 weeks
Team
6 engineers
Interactive 3D · drag to rotate
−30%
forecast error
−34%
stock-outs
78k
SKUs modelled

The challenge

Conditions on site — heat, dust, vibration and patchy connectivity — had defeated two earlier attempts at this in retail.

What we built

We owned the whole stack — hardware, firmware, models and dashboards — using LightGBM, Prophet, PostgreSQL so there was one accountable team.

Delivery timeline

  1. Phase 1

    Discovery sprint and site survey

  2. Phase 2

    Prototype built and tested in the Motera lab

  3. Phase 3

    Pilot on site with before/after measurement

  4. Phase 4

    Scale-out, training and handover

Highlights

  • Operator training delivered in Gujarati, Hindi and English
  • Designed, built and supported from Motera, Ahmedabad
  • Before-and-after measurements agreed with the client up front
  • Documentation, drawings and source handed over at close

Have a problem that needs a machine built for it?

Tell us the constraint — line speed, accuracy, environment, deadline. An engineer will read it and reply with a straight answer, including when the answer is that you don't need us.

We reply to every enquiry within one business day.