Bottom-Up Hierarchical Forecasting

Forecasting Product Demand in R

Aric LaBarr, Ph.D.

Senior Data Scientist, Elder Research

Hierarchical Forecasting

  • Hierarchical forecasting can be used when different items that need to be forecasted can be arranged in a logical hierarchy
  • Forecasts need to be reconciled up and down the hierarchy
Forecasting Product Demand in R

Hierarchical Structure

Forecasting Product Demand in R

Types of Hierarchical Forecasting

  • Three Types of Hierarchical Forecasting
    1. Bottom-up Forecasting
    2. Top-down Forecasting
    3. Middle-out Forecasting
Forecasting Product Demand in R

Bottom-up Forecasting

Forecasting Product Demand in R

Bottom-up Forecasting Example

for_M_total <- pred_M_hi_xts + pred_M_lo_xts

M_t_v <- bev_xts_valid[,"M.hi"] + bev_xts_valid[,"M.lo"]

MAPE <- 100*mean(abs((for_M_total - M_t_v)/M_t_v))
print(MAPE)

[1] 7.599677
Forecasting Product Demand in R

Let's practice!

Forecasting Product Demand in R

Preparing Video For Download...