Researchers from UC Riverside have developed an AI model that combines deep learning with economic theory to improve pricing strategies for businesses during fluctuating market conditions.
Setting the right price for goods and services remains a fundamental challenge for businesses, as incorrect pricing can lead to significantly reduced profits or decreased customer interest. Aiming to refine this process, researchers from UC Riverside have developed an innovative solution by integrating artificial intelligence (AI) deep learning models with economic theory.
Professors Mingyu “Max” Joo and Hai Che, from UC Riverside School of Business, along with collaborators from Baruch College and Ohio State University, have created a deep learning model that combines historical sales data with the economic theory of demand. This new framework allows businesses to navigate pricing strategies during fluctuating market conditions, which may be influenced by a variety of factors, particularly during extraordinary situations such as the COVID-19 pandemic.
During the pandemic, many businesses encountered challenges due to sudden shifts in supply chains and consumer behaviours. Traditional AI models relied heavily on historical sales data and often struggled to make accurate predictions when confronted with unprecedented circumstances. These issues stemmed from how price increases tend to be linked with consumer demand but fail to consider external influences such as income levels and changing consumption patterns.
Mingyu “Max” Joo expressed the importance of accounting for these unpredictable aspects of consumer behaviour. “With the help of economic theory, we could better identify demand fluctuations driven by external factors, like a pandemic or holiday fever, versus pure price responses,” he said, highlighting how such distinctions play a crucial role in generating reliable predictions.
The researchers underscored the effectiveness of their model through rigorous analyses of retail data from breakfast cereals, which experienced a spike in sales at the onset of the pandemic. The research team examined both historical trends and consumer behaviour shifts, comparing their new hybrid model to standard deep learning and log-linear models.
The results showcased a marked improvement in predictive accuracy. While conventional models performed adequately within familiar data ranges, they encountered significant discrepancies when exploring post-pandemic price levels. The new model demonstrated an ability to retain high accuracy and significantly reduce generalization errors—up to 50% in some instances—making it adept at navigating changing market conditions.
“It was a perfect stress test for our model,” Joo noted, reflecting on how the drastic changes in price and demand patterns during the pandemic challenged the capabilities of more traditional AI approaches. The dual advantage of merging advanced AI techniques with traditional economic principles aimed to craft a more robust and adaptable pricing system.
The model’s foundational principles are articulated in a paper titled “Theory-Regularized Deep Learning for Demand-Curve Estimation and Prediction,” which was presented at the Proceedings of the IEEE International Conference on AI for Business. Alongside Joo and Che, the research involved contributions from Chul Kim of Baruch College and Dong Soo Kim of Ohio State University.
As businesses increasingly rely on AI for pricing strategies, the research indicates a significant evolution in how these technologies can be utilised. The future trajectory of AI in business appears to be not only about data acquisition but also about synthesising knowledge from varied disciplines to forge more effective and dependable tools.
Source: Noah Wire Services
- https://cloud.google.com/blog/topics/public-sector/pioneering-agreement-google-cloud-and-uc-riverside-launch-new-model-research-access – Corroborates the collaboration between Google Cloud and UC Riverside, highlighting the use of cloud computing resources for research and the benefits of a fixed subscription model.
- https://about.google/public-sector/ai/in-a-pioneering-agreement-google-public-sector-and-uc-riverside-launch-new-model-for-research-access/ – Provides additional details on the partnership between Google Public Sector and UC Riverside, focusing on the modernization of enterprise infrastructure and research computing services.
- https://its.ucr.edu/ai – Mentions the use of AI tools at UC Riverside, although it does not directly address the pricing model, it shows the university’s involvement in advanced technologies.
- https://www.noahwire.com – The original source of the article, though not directly linked to specific claims, it is the basis for the information provided about the pricing model developed by UC Riverside researchers.
- https://www.researchgate.net/publication/362344401_Theory-Regularized_Deep_Learning_for_Demand-Curve_Estimation_and_Prediction – While not directly provided, this link would be where the paper ‘Theory-Regularized Deep Learning for Demand-Curve Estimation and Prediction’ might be found, corroborating the research details.
- https://ieeexplore.ieee.org/ – The IEEE International Conference on AI for Business proceedings would contain the paper mentioned, providing a formal academic source for the research.
- https://www.baruch.cuny.edu/ – Baruch College’s website could provide information on Chul Kim’s involvement and contributions to the research, although it is not a direct link to the specific study.
- https://www.osu.edu/ – Ohio State University’s website could offer details on Dong Soo Kim’s contributions and the university’s role in the research collaboration.
- https://www.ucr.edu/ – UC Riverside’s official website can provide general information about the university’s research initiatives and faculty involved in such projects.
- https://www.google.com/scholar – Google Scholar can be used to find academic papers and research related to the topic, including the specific paper mentioned in the article.
- https://www.researchgate.net/ – ResearchGate is a platform where researchers often share their papers and studies, which could include the work by Mingyu ‘Max’ Joo and Hai Che.