Download PDFOpen PDF in browser

Using Local Weather and Machine Learning to Forecast Market’s Demand and Supply: A Pilot Study (November 2019)

EasyChair Preprint no. 4520

12 pagesDate: November 7, 2020

Abstract

The study explored Long-Short-Term-Memory (LSTM) neural networks (ANN) to predict local Small and Medium Scale Entrepreneurs’ (SMEs’) goods purchases quantities, buying and selling prices. The independent variables included low and high temperatures. The study was exploring the impact of using ANN models in forecasting local goods’ quantity purchase, buying prices, and the goods selling prices, from the day-to-day local weather conditions, on the local SMEs. The weather was the independent variable for forecasting. The study’s SMEs included marketers trading in household edible commodities, like bread, mealie-meal, cooking oil, green vegetables and sugar. LSTM modular was employed to forecast the daily goods’ quantity purchase, buying and selling prices. It was observed that ANN can use the weather to forecast demand and supply parameters for market traded goods. The study used the weather in the LSTM ANN model to forecast demand and supply to establish the feasibility to build an application embedded with ANN model which can be used by the local SMEs for insisting into their businesses’ future. This will help SMEs plan for their localised markets, to combat unprecedented demand-and-supply fluctuations, inflation adversities, price instability, excess inventory, etc

Keyphrases: Artificial Neural Networks, Demand Forecasting, Forecasting, machine learning, market research, supply and demand, Temperature

BibTeX entry
BibTeX does not have the right entry for preprints. This is a hack for producing the correct reference:
@Booklet{EasyChair:4520,
  author = {Jephter Pelekamoyo},
  title = {Using Local Weather and Machine Learning to Forecast Market’s Demand and Supply: A Pilot Study (November 2019)},
  howpublished = {EasyChair Preprint no. 4520},

  year = {EasyChair, 2020}}
Download PDFOpen PDF in browser