Use case: synAPs
Sales Budget Predictive Model
Organizations involved
A multinational company in the chemical sector with global production and logistics offices and Studi di Sicurezza S.n.c. (innovative startup, solution provider)
Objectives obtained
The project has produced:
- a predictive model for the sales budget, analyzing the historical data of the last five years;;
- a model that makes it possible to obtain sales budget forecasts in a very short time, being able to make comparisons with turnover and providing any deviations;;
- a model that makes it possible to evaluate the trends in sales budgets divided by Line through graphs;
- the possibility for the company to have a tool that allows it to significantly optimize the sales budget process.
The challenge
Financial time series are non-stationary and non-linear data that are influenced by external factors. There are several performing predictive approaches such as the ARIMA model and Exponential Smoothing. Accurate forecasting of budget data is a strategic and challenging task for optimal resource management, requiring the use of the most accurate model. Forecasting sales budget data is different from typical predictive and machine learning applications, as it is not about just replicating the tasks that humans can easily do.
The solution
YoctoIT proposed a predictive approach that uses and compares the ARIMA model of Machine Learning and the Exponential Smoothing model. The application and benchmarking were able to demonstrate that the Exponential Smoothing model outperforms the ARIMA model, mainly in terms of forecast accuracy and computation speed.
In 40 days of analysis and development, the team made up of 1 marketing expert (employed 10% of the time), 1 Data Engineer (10%), 1 Data Scientist (70%), 1 Machine Learning Engineer (10%) , allowed to:
- analyze the client’s organizational model;
- define the correct procedures for accessing the various data sources;
- improve the quality, completeness and granularity of data;
- develop the prototype (proof of concept);
- evaluate and validate the model obtained with the client;
- test the solution in the field and make it permanently available.
The project produced a predictive model for the sales budget, analyzing historical data from the last 5 years.
The model allows you to obtain sales budget forecasts in a very short time, being able to make comparisons with turnover and providing any deviations.
The model allows you to evaluate the sales budget trends divided by Line through graphs.
The company now has a tool that allows it to significantly optimize the sales budget process.
Download here the Pdf slides describing the project, while clicking here you can play with a demonstration in real time.