Data-driven Research On Competency Model Of Managers In The Enterprise

The traditional competency model lacks objective and quantitative criteria,and the construction methods mostly rely on personal experience and subjective judgment,which fails to effectively play the value of human resource data.Therefore,this study will conduct research on data-driven competency model of managers in the enterprise,take a state-owned enterprise as an example,select managers who have served as a middle management position as the research object,and quantify the competency model based on human resource data.The main work and achievements of this paper are as follows:(1)Design the personnel description indicator systemBased on the existing human resource data,a set of personnel description indicator system including 51 indicators is designed from basic information dimension,work information dimension and study information dimension.We extract and analyze the indicator data of a group of middle-level managers,which provides data support for building the competency model.(2)Construct the competency model of middle-level managers in the enterpriseBased on the indicator data of middle-level managers,we construct the competency model of middle-level managers in accordance with the process of performance indicator determination,calibration sample selection,index analysis,index induction and model determination.The competency model describes five competencies of excellent middle managers,respectively are the richness of work experience,learning ability,management ability,work specificity and self-discipline ability.(3)Verify the competency prediction modelIn order to verify the competency model,machine learning classification algorithms are used to build the competency prediction model.Results show that the Adaboost model has better prediction performance,and the competency indicator data can effectively assess whether employees are competent for the management position,which proves that the data-driven research method of competency model proposed in this paper is scientific and effective.