Economic Network

• It is a type of TensorFlow based network for identifying the relationships of different factors on overall sales.
• It acts as the parent network to BFNet and TemporalNet and contains common implementations shared by both TemporalNet and BFNet.
• TemporalNet and BFNet then calculates the baseline separately.

Objective

• To establish the relationship of different investments & other exogenous factors (temperature, precipitation, macroeconomic factors, etc.) with overall sales that helps in sales prediction.

• To determining the contribution of investments (including Media, traditional, sponsorship, price discount, etc. spending) in the uplift of sales (incremental revenue) by generating diminishing return curves that helps in determining the ROI.

Input Data

• Historical Data (from 2018 till date) related to

  • Sales - @State & brand level
  • Investments - @State, brand, investment vehicles (includes Media, trade, etc.)
  • Weather - @State level
  • Macroeconomic factors - @State level
  • Pricing - @State & brand level

Output

Given the objective, the output is also 2-fold: -

• Sales prediction (at desired granularity)
• ROI curves for investment vehicles like Facebook, ENT TV, Promo Discounts and so on (at desired granularity)