Analysis and forecasting of key professions' profitability to prevent labor market oversaturation in Kazakhstan

The project focuses on analyzing and forecasting the profitability of key professions in Kazakhstan using machine learning methods to prevent labor market oversaturation. The core idea is to develop a model that identifies professions at high risk of income stagnation and unemployment, enabling the redistribution of labor resources and fostering the development of less saturated but promising industries.

Problem
Kazakhstan's labor market is currently experiencing an imbalance: the growing number of educational grants, particularly in IT and popular specialties, is not accompanied by a corresponding increase in job opportunities. This leads to declining profitability of professions, heightened competition, and rising unemployment, especially among young people. At the same time, sectors like agriculture and trades face a persistent shortage of workers.

Solution
The use of the CatBoost algorithm enables precise forecasting of changes in profession profitability up to 2030. The project analyzes data on income, education level, age, and professional affiliation to propose measures for redistributing labor resources and supporting promising industries.

Key Advantages
- Accurate forecasting of profession profitability and market risks.
- Support for sustainable development through resource redistribution and stimulation of emerging industries.
- Innovative application of machine learning methods for big data analysis.
- Integration potential into government employment strategies and educational initiatives.
- Long-term planning capabilities extending to 2030.

Impact
The project combines innovative approaches in big data with a focus on sustainable development goals, offering solutions to improve economic stability and align education with labor market demands.

CatBoost algorithm: a gradient boosting method optimized for handling categorical data.

PowerTransformer (Box-Cox): a tool for normalizing the distribution of the target variable.

Label encoding: a method for encoding categorical data such as territorial and occupational codes.

The model was trained on a feature matrix \( X \), which included the following independent variables:

- SUO: Level of education
- Age: Age
- SPPN2: Employment start date (year)
- KATO: Territorial code
- GSDPRS: Occupational classifier

The target variable \( y \) represents income (\( VAL\_NUM \)), transformed using the Box-Cox method to stabilize the distribution and enhance prediction accuracy.

Concept

Public administration:
Optimization of employment policies and labor resource allocation.
Economic planning:
Forecasting changes in profession profitability to ensure sustainable labor market development.
Education:
Designing training programs to prepare specialists for in-demand industries.
Social policy:
Reducing unemployment rates and supporting vulnerable population groups.
Private sector:
Analyzing labor market trends to optimize hiring and business planning.
Science and research:
Studying socio-economic factors influencing the labor market.

The project successfully implemented the analysis and forecasting of the profitability of key professions using advanced machine learning methods, such as CatBoost. This enabled:

Development of a functional forecasting model capable of accurately predicting changes in profitability up to 2030.
Identification of key labor market issues, including professions at high risk of oversaturation, along with proposed measures to address them.
Improvement in data analysis quality through tools for distribution normalization and categorical data processing.
Potential Impact on Society and Business
For Society:

The project can help reduce unemployment risks and increase employment rates in promising industries.
Its findings can enhance the allocation of educational grants by aligning them with real market needs.
For Business:

Companies can utilize forecasts for more effective workforce planning and retention strategies.

Economic impact
Potential cost reduction:
The forecasting model enables the government to cut expenditures on unemployment support programs by early identification of professions at risk of oversaturation. For instance, reallocating grants to less saturated specialties can reduce costs associated with subsidies and retraining.
Increased economic efficiency:
Accurate predictions help businesses optimize hiring processes and minimize expenses related to training and onboarding new employees.
Social impact:

Reduction of unemployment:
The project's recommendations can help decrease unemployment rates by redistributing labor resources to less saturated but promising industries.
Improvement of the educational system:
Optimizing grant allocation and creating new training programs based on forecasts ensures better alignment with market demands.
Income stabilization:
Supporting professions at risk of income stagnation helps prevent a decline in the standard of living for workers in these fields.

Unique features of the Project

Use of CatBoost

The application of the algorithm developed by Yandex enables the processing of large volumes of categorical data with high prediction accuracy. This approach is rarely employed in labor market forecasting tasks.
Focus on local data

The project accounts for the unique characteristics of Kazakhstan's labor market, including regional disparities, demographic factors, and education levels, making it well-suited to local realities.
Long-term forecasting

Forecasting profession profitability through 2030 allows for planning changes in the education system and labor policies while considering long-term trends.
Integration of economic and social analysis

The project goes beyond income data analysis by examining its impact on social aspects such as employment, competition, and labor resource distribution.
Flexibility and adaptability

The model is highly adaptable for use across various industries and can be customized to meet the needs of specific clients, whether governmental agencies or private businesses.
Practical applicability
The project provides specific recommendations for redistributing labor resources and fostering promising industries, making it not only an analytical tool but also a strategically valuable resource.

Project scalability
Geographic expansion

Adapting the model to analyze labor markets in other Central Asian countries or regions with similar economic and social conditions.
Integration of additional data

Incorporating data on international migration, changes in tax policies, and investments in high-tech industries to enhance the accuracy of forecasts.
Implementation in public and private management systems

Leveraging the project to develop national employment strategies and educational programs tailored to address labor market challenges effectively.

The dataset for this project was provided by the Bureau of National Statistics under the Agency for Strategic Planning and Reforms of the Republic of Kazakhstan.

It was awarded by the Agency for Strategic Planning and Reforms at the National Competition for the application of Big Data to address the country's socio-economic and methodological challenges.