The Historical Roots of Forecasting
Since the beginning of human history, people have sought to predict the future. Ancient practices like BaZi, oracle bone divination, and lot-drawing in China, alongside Western methods such as tarot reading, show how deeply this desire is ingrained in human culture. Across many societies, interpreting dreams to predict future events was also widespread. In China, the Song Dynasty’s Taishi Bureau used early meteorological principles to forecast weather, supporting agriculture and other activities. Similarly, the Qintianjian in the Ming Dynasty focused on astronomy and divination for significant events. Some of these approaches share similarities with modern science, while others remain rooted in mysticism.
However, the desire to forecast was not always welcomed. In some periods, it was seen as a challenge to authority.
Forecasters had a tougher time under the emperor Constantius II, who issued a decree in AD357 forbidding anyone “to consult a soothsayer, a mathematician, or a forecaster … May curiosity to foretell the future be silenced forever.” 1 A similar ban on forecasting occurred in England in 1824 2 when “every person pretending or professing to tell fortunes” was “deemed a rogue and vagabond”. The punishment was up to three months’ imprisonment with hard labour!
In contrast, modern forecasting is vastly different. Today, it is a scientific tool applied across many areas, such as business, economics, politics, traffic and environmental planning. Accurate forecasts are essential for making informed decisions, guiding strategies, and managing resources. No longer dismissed as mystical or forbidden, forecasting has evolved into a sophisticated field powered by mathematics and advanced technology.
At its core, humanity’s fascination with forecasting stems from an intrinsic drive for knowledge, security, control, and prosperity.
Modern Forecasting
With the development of computers, we can now analyze massive datasets using advanced algorithms, achieving levels of precision once thought impossible. Early computer-based forecasting relied on structured models like linear regression and decision trees, which mathematically described how systems behaved. Over time, forecasting has shifted to methods such as deep learning and neural networks. These tools can uncover hidden patterns in vast datasets but often act as “black boxes,” making their processes difficult to understand.
In my view, forecasts generated by these opaque systems sometimes feel as mysterious as gazing into a crystal ball. This lack of transparency creates challenges in trust and usability, especially in critical areas like medical diagnosis, self-driving cars, or financial planning. To address this, we must prioritize not only accuracy but also interpretability. Transparent and understandable forecasting gives decision-makers confidence and helps avoid unintended consequences.
A Journey Into Forecasting
In this series, I will follow the work of Rob J. Hyndman and George Athanasopoulos, exploring traditional forecasting methods through their book, Forecasting: Principles and Practice 3. Through this journey, I aim to gain insights into time series forecasting and the importance of interpretability, contributing to a broader understanding of this essential field.