Decision support of the unemployment fund OTT

The unemployment fund's decision support OTT is a data-based tool that predicts the probability of an unemployed person moving to work, identifies the factors that influence it, and thereby helps to provide assistance to the unemployed based on individual needs and increase the effectiveness of the unemployment fund.

Approximately 70,000 people register as unemployed each year, all of whom have different backgrounds, strengths and obstacles. The 350 counselors of the unemployment fund have to analyze a large amount of information in order to make a suitable plan together with the person to help him find work. This process takes time, but it is important to understand the person's situation as soon as possible and start acting in the right direction. Using a machine learning model trained on the basis of the unemployed data of the previous five years, OTT summarizes the situation of a specific person, predicting the probability of his moving to work during the year, the probability of being unemployed again and highlighting the circumstances that affect it. In this way, the advisor gets a quick summary of his client's situation, as well as an overview of all his clients based on the OTT assessment, and this allows him to set priorities according to the extent of the client's need for help.

In addition, OTT assessments are also used by advisor managers, who can more evenly distribute the workload of advisors and support advisors with more complex client portfolios.

The analytical side of OTT runs in R software and uses a random forest model. OTT is integrated with the reporting module (data warehouse) of the unemployment fund, through which the data Ri used in the model and the predicted outputs move to the adviser's desktop.

As the labor market situation changes rapidly, it is also necessary to keep OTT constantly up-to-date. To this end, advisers using OTT can give feedback on a daily basis, and based on this, the model is improved by adding new features or removing non-functional ones. Also, the model needs to be retrained once a quarter when data is added.

OTT has been created in cooperation with the unemployment fund and development partners CITIS, Nortal and Resta.

In summary, OTT helps to improve the quality of the offered service and increase efficiency by reusing the data generated during the basic process of the unemployment fund.