Forecasting leadtime and demand

The script below illustrates how probabilistic forecasts can be produced, first for leadtime, second for demand. The forecasts are persisted to a Ionic file. The Ionic file format is required because regular flat files - e.g. CSV - don’t support the direct export of the distribution data type. Finally, the distributions are extended into a grid.

/// Probabilistic forecasts both for leadtime and demand
read "/sample/Lokad_Items.tsv" as Items [Id] with
  Id : text
  Category : text
  SubCategory : text
  Brand : text
  Supplier : text
  LeadTime : number

read "/sample/Lokad_Orders.tsv" as Orders expect [Id, Date] with
  Id : text
  Date : date
  Quantity : number

read "/sample/Lokad_PurchaseOrders.tsv" as PO expect [Id, Date] with
  Id : text
  Date : date
  DeliveryDate : date
  Supplier : text

path = "/sample/"

// Forecasting lead time distribution with purchase orders history
Items.Leadtime = forecast.leadtime(
  category: Items.Brand, Items.Category, Items.SubCategory
  supplier: Items.Supplier
  offset: 0
  present: (max(Orders.Date) by 1) + 1
  leadtimeDate: PO.Date
  leadtimeValue: PO.DeliveryDate - PO.Date + 1
  leadtimeSupplier: PO.Supplier)

// Forecasting demand using varying lead times and sales history
// (in practice, the ordering leadtime needs to be factored in as well)
Items.Demand = forecast.demand(
  category: Items.Brand, Items.Category, Items.SubCategory
  horizon: Items.Leadtime
  offset: 0
  present: (max(Orders.Date) by 1) + 1
  demandDate: Orders.Date
  demandValue: Orders.Quantity)

// Persisting the ranvars into a Ionic data file
show table "Ranvars" write:"\{path}Lokad_Ranvars.ion" with Id, Demand

//Extending the demand ranvar into a grid
table Grid = extend.ranvar(Items.Demand)
Grid.Probability = int(Items.Demand, Grid.Min, Grid.Max)

show table "Grid" with
  Id
  Grid.Min as "Min"
  Grid.Max as "Max"
  Grid.Probability as "Probability"