Market Anomaly Detection through Data Science – Dr. Ivan Abakumov draws the line from Theodicy to Market Anomaly Detection. Meet HQ revenue's Lead Data Scientist.
I am the Lead Data Scientist at HQ revenue, where my focus is on using best research practices, state-of-the-art machine learning, and high-performance computing to uncover insights in the hospitality industry.
My educational background is in Physics, with a specialization in Geophysics. During my studies, I was fortunate to work on several research projects with Shell, collaborating with some of the top minds in the industry.
In 2016, shortly before defending my Ph.D. thesis, I read a paper by Brendon Hall that introduced machine learning concepts for facies classification. This paper was a pivotal moment for me, and I knew then that I wanted to pursue machine learning in my future work.
At the time, it was challenging to find a postdoc position focused on machine learning, but I was fortunate to secure a position at FU Berlin. In this role, I was able to teach courses on machine learning and deep learning to both master's and Ph.D. students in geophysics. I also collaborated on an educational project with DGMK to provide industry professionals with machine learning training. Despite my theoretical knowledge, I was eager to apply my skills to real-world business problems. That led me to my current position at HQ revenue.
My job is to leverage machine learning techniques to gain insights in the hospitality industry. Specifically, we work with a large amount of historical and forward-looking data to develop demand and occupancy forecasts for our customers. Additionally, we monitor thousands of markets daily and provide analytics and trends to our clients.
The anomaly project is a natural extension of our work. Since our clients receive a significant amount of information from us, it can be challenging for them to keep track of pricing, demand and occupancy developments on a day-to-day basis. The anomaly project helps us highlight the most important days for our customers, making it easier for them to stay up-to-date with changes and make informed decisions.
One of the most challenging aspects of working on the anomaly project was creating individual models for each customer. It's amazing to see how hotels differ worldwide and the varying behaviours we see in different markets.
I'm proud that we were able to formulate all of our algorithms in such a way that there are no predetermined parameters - each customer receives a model that is purely tuned based on their own unique historical data.
Currently, we offer our customers six basic anomalies that cover the most important KPIs, such as changes in demand, property occupancy, and key changes in prices.
However, I believe there is potential for many more anomalies to help our customers gain even deeper insights into their businesses. I would love to hear feedback from our customers on the types of anomalies they would like to see.
It is widely recognized that data is the new oil. Although my previous career in the oil and gas industry may seem disconnected from my current work in the hotel industry, there are valuable skills that I can apply to any field. My background in research has equipped me with abilities such as data analysis, problem-solving, and attention to detail. Furthermore, my extensive business travel experience has given me a unique perspective on what constitutes an exceptional hotel experience for guests. These skills and experiences have proven to be incredibly beneficial in my day-to-day work.
I really like to repeat the following question from Terry Pratchett's Night Watch: "Is every accident just a higher-order design?"
As data scientists, we are often able to discern patterns in the data that explain what may seem like random fluctuations in the market - a "higher-order design."