Page 14 - AeM_April_2022

RESEARCH, ANALYSIS & TRENDS

Novel AI-powered forecasting models

needed as many fail in times of crisis

must be retrained with the current data material
and any new influencing variables must be identi-
fied and modeled. In addition, deep learning can be
used to recognize new patterns, such as neural
networks, and recognized patterns can in turn be
used for the forecast. To be on the safe side, those
responsible should also consider so-called ensem-
ble models which are composed of a large number
of different AI models, which means that one-sided
adjustments to the individual models can be aver-
aged out and lead to more robust forecasts.

As adesso said right at the beginning of the Corona There are opportunities to adapt existing models to the
crisis, the stability of long-term forecasts has been current situation, and thus protect them from becoming
completely turned upside down since the pandemic obsolete. But what happens when the next concept drift
and has led to high volatility in forecasting trends is due? This is already announced with the lifting of
and planning uncertainty for companies. current restrictions and shows how crucial it will be to
be able to react to changes and the associated deterio-
In fact, many forecasting models are failing in the cur- rations of the model in the future.
rent crisis situation, which is why it is high time to take
active countermeasures. Where in "normal times" deci- Especially in times when the much-cited data-driven
sion-makers derived action-guiding forecasts from the company is increasingly serving as a vision for the fu-
AI analysis of large amounts of data, companies are ture of one's own company, the importance of regulated
struggling with completely changed framework condi- processes in data science operations is increasing.
tions. New operating concepts for AI factories are establish-
ing themselves here, in which well-known approaches
Being able to predict customer behavior, the hotline or from software development, such as agile develop-
machine utilization, today has two main reasons. ment, CI/CD or DevOps, are becoming increasingly
important.
• Firstly, the data situation has changed radically in
terms of quantity and quality. Analysts speak of Expert knowledge is indispensable in order to generate
concept drifts that lead to the formation of new pat- new, perspectively valuable connections from the
terns in the data set and bring existing models changed amount of information. In addition to expertise
closer to their expiration date. Recognizing the cur- in AI and data science, industry knowledge and speed
rent radical distortions in customer behavior is the are required to provide companies with the much-
task that companies must now solve as quickly as needed, well-founded information for rapid decision
possible. This applies, for example, to customer support in uncertain times. ◊
sales at banks, call analysis at authorities and call
centers or fault forecasts for IT support or technical By Daniela La Marca
systems due to changed workloads.

• Secondly, this extreme imbalance in the data has
rendered many forecasting models obsolete. It is
therefore necessary to adapt the AI models to the
changed data situation, because despite the ex-
ceptional situation, companies can continue to
make forecasts if the experts can quickly make the
appropriate adjustments. To do this, the models

14 April 2022: Marketing Automation: AI, Big Data & Deep Learning