/pibic_2021-2022

Complete solution for MOFC M5 Forecasting in kaggle.

Primary LanguagePythonMIT LicenseMIT

Institutional Scientific Initiation Scholarship Program 2021-2022

This repository documents my participation in the 2021 Institutional Scientific Initiation Scholarship Program at the University of Brasília (UnB) PIBIC. The program lasted for one year and focused on Time Series Analysis. My research involved working with ARIMA models with explanatory variables, Dynamic Harmonic Regression, and TBATS models under the guidance of Prof. Dr. José Augusto Fiorucci from the Department of Statistics (EST/UnB).

Overview

This project presents a comprehensive solution for the 2020 MOFC M5 Forecasting competition on Kaggle. We forecasted the daily sales of retail products sold by Walmart in the U.S., using explanatory variables such as price, payment method, and commemorative dates provided by Walmart for the competition. The aim was not only to generate accurate forecasts but also to submit our solution to the M5 Forecasting competition to benchmark our results against those of other participants worldwide.