Learn how the COVID-19 crisis challenged algorithms based on historical data and how companies adapted to the unforeseen. Solutions include active learning, retraining systems, and the importance of exceptional data for enhancing the resilience of predictive models in the face of global crises.
Data and Algorithms in the Face of the Unexpected: Adapting Systems to Crises and Enhancing Resilience
The COVID-19 crisis has exposed the limitations of systems relying on historical data and algorithms trained on past events. Predictive models, especially in stock management and consumer behavior, have become ineffective when faced with the unexpected. Businesses had to adapt quickly, sometimes relying on human intuition to make crucial decisions.
The rise of machine learning and big data has revolutionized business operations. However, in the face of events like a pandemic, algorithms trained on past data no longer serve their purpose. For example, sales and stock prediction models, designed for normal circumstances, became ineffective when consumer habits radically shifted.
To overcome these challenges, various strategies are being proposed. Some businesses retrained their algorithms with newer data, while others opted for active learning, allowing intelligent systems to quickly adapt to new information. One key lesson is the need to react swiftly to the unexpected and to increase the frequency of model training to make systems more agile.
COVID-19 has also highlighted the value of historical data, often underestimated. While past data might seem outdated, it can offer valuable insight into future crises. The importance of preserving such data is thus becoming crucial, even during times of change. The lessons learned from the pandemic could reshape how businesses approach future crises.
Source : ICTjournal
