/Time-Series-Forecasting

A time series project using Arima

Primary LanguageJupyter Notebook

Time-Series-Forecasting

1.1 Business Overview

Sales prediction has become the new norm in most medium to large-sized stores. It is a crucial business exercise that allows forecasting of future sales revenue. It is based on historical data, industry trends, and the status of the current sales pipeline. Through accurate sales forecasting, sales managers can make smarter decisions about things that affect cash flow, for example, budgeting, hiring, or goal setting. On the flip side, inaccurate sales forecasts leave the sales managers guessing whether the predicted sales will actually get to the quota. This could result in the sales managers not being aware of any problems in the sales pipeline in time to fix them.

In this study, we will be predicting the sales of Favorita Stores in Ecuador, whose largest population is Latin Americans. Ecuador is a major petroleum exporter, therefore oil prices directly affect product prices. It is also prone to earthquakes, volcanic eruptions, and tsunamis. The most recent major earthquake was on April 16th, 2016. The most recent was 2022-03-26 which had minor damage.

1.2 Problem Statement

Favorita Stores in Ecuador has thousands of product families such as automotive, baby care, beauty, beverages, books, etcetera. The sales managers have had a hard time controlling sales and stock inventory due to poor sales predictions in the past. The inaccurate predictions usually hit the inventory the hardest because failing to anticipate surges or troughs in customer demand can lead to undersupply or oversupply in inventory, both of which can have negative consequences. An undersupply of products erodes customers’ confidence, reduces profits, and hands a golden opportunity to competitors to fill the gap in the market. An oversupply increases inventory costs, as well as creating an imbalance between the cost of production and sales receipts. Either way, inventory problems caused by poor forecasting can seriously affect a business’s cash flow and profit margins.

The impact of the unbalanced natural phenomena is especially felt in oil production and pricing, which in turn affects sales. To reduce the impact of the negative consequences, the sales managers, through a team of data scientists, decided to make a prediction of sales by predicting oil price fluctuations so that Favorita Stores could prepare accordingly for the expected oil price fluctuations caused by the earthquakes.