The Endless Possibilities of Forecasting in Data Science

Author(s): Alexandre Warembourg

Originally published on Towards AI.

Discover the numerous methods available for forecasting in data science through practical examples

When I first started my journey in data science, my initial task was on forecasting. At the same time, I had just completed my Master’s degree in econometrics. My first impression of forecasting was rather dull and monotonous, as I viewed everything through the prism of time series econometrics, which involved plotting partial autocorrelation and autocorrelation plots to manually determine the correct parameters of AR and MA for defining an ARIMA model. However, I now realize that this was an incomplete perspective of the reality of statistical forecasting, as I was a novice in many ways.

After several successful forecasting projects, I have learned that the field of forecasting differs significantly from classical regression problems and can be approached in various ways beyond statistical predictions. This expands the possibilities of modeling when beginning a project.

Let’s examine the multitude of forecasting options available through the prism of the Favorita grocery sales forecasting competition on https://www.kaggle.com/c/favorita-grocery-sales-forecasting/overview. This involves predicting sales for various store-product combinations 16 days in advance.

I will not do an in-depth analysis because we will only use a subset of the training data. The data follows a standard structure, including store ID, item ID, unit sales, day, and a promotion flag.

Source… Read the full blog for free on Medium.

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI