Workperf-An intelligent Predictive model for work performance rating .pptx

Modelling the control and automation of digital devices provides a great deal of knowledge about the working principles of electronic gadgets and devices. It also aids in the development of prototypes and helps determine their efficiency and accuracy ratings before or when made available for use.

1. WORK-PERF: An Intelligent Predictive Model for Work Performance Rating Department of Computer Science, Federal University of Agriculture, Abeokuta, Nigeria . Femi Johnson , Saidat A. Onashoga , Ibharalu F. Thomas , Opakunle Victor and Adenusi Cecilia Thursday, 23rd February 2023.

2. Presentation Outline 01. Introduction 02. Related Works 03. Methodology 04. Result Analysis 05. Findings and Conclusion References

3. 01 Introduction WORK-PERF: An Intelligent Predictive Model for Work Performance Rating

4. Introduction Modelling the control and automation of digital devices provides a great deal of knowledge about the working principles of electronic gadgets and devices. It also aids in the development of prototypes and helps determine their efficiency and accuracy ratings before or when made available for use [28]. The United Nations Population Fund (UNPF) in 2022, estimated the world's population at about 7.8 billion with China possessing about 1.42 billion people. A similar data provided by the World's review (an online database) on the percentage of the female is also estimated at 3.97 billion representing 49.59% of the world's population. Considering the significant roles played by the female gender [29] in the family, society, and the world at large, it's of paramount interest that adequate provisions are made for their education, security, mental and physical well-being [15] in support of the United Nations' effort. The research conducted in this presented paper unravels the untruthful myth about the low work performance ratings often attributed to the female gender through the development an Artificial Intelligence (A.I) model to automatically predict psychological health status as a major determinant of work performance with accuracy comparison check on similar machine learning models as listed by [26]. 01

5. 02 Review of Literature

6. Review of Related Works The modeling and integration of psychological factors as predictors of the work performance ratings of selected employees with high school certificates to Master's degree holders in five major departments in an organization [21] provided innovative means required by managers at different levels to help create conducive and friendly work environments for employees. Jean’s work performance model deployed to test the effects of nine factors on teachers in Nord-Pas de Calais area of France indicated that the availability of resources for teaching provides meaningful and enjoyable social environments for staff [12]. Job roles are segregated by gender [1] and men are more preferred in physical related tasks like farming, and construction while women are considered more reliable in the education, health, domestic, and other related care sectors. 02 In addition, a lower risk of job loss favours women in the care sector (health) than men . Generally, women are on average at a disadvantaged level of job loss risk [11, 18]. A research conducted on the risk of automation in Latin America revealed the gender [22] at a higher risk of job loss. The less educated are also at higher risks of displacement with the adoption of technology than the highly educated ones [8, 6]. There has been a continuous increase in the number of female participation in the maritime industry [5] but barriers such as physical, social and psychological challenges are still hindering their full involvement [10]. Social demographic factors significantly contribute to organizational commitment and work-life balance [16] among the female gender working in the hospitality industries. Work and time support are the major predictors for determining work-life commitment.

7. Methodology 03

8. Work-Perf development Stages Data collection and pre-processing . Online Data source with selected metrics. Similar but varied number of data samples at both phases . The Work-PERF’s model development. AI technique on PCA reduced features . A multi-components based machine learning evaluation criteria . Model training and testing. Work-PERF Model evaluation. 03

9. Data Pre- processing Work-Perf development Stages 03 Two groups of datasets containing equal number of data samples (400) were later supplied to the model for testing and validation based on the knowledge it has gained at the training phase with the inference rules stored in its inference rule engine. The data was normalize and re-populated with random data through Synthetic Minority Oversampling (SMOTE) analysis to reduce overfitting. The pre-processed dataset comprising of about 2400 data samples was divided into 50%. 25% and 25% for training, testing and validation respectively. Model training and testing. V ariable modeling for Work-Perf

10. Work-Perf Framework 03

11. Presentation Progress Introduction Review of literatures Methodology 20% 40% 60%

12. Result Analysis Implementation Result Analysis R esult Comparision Evaluation 4

13. 04 Work-Perf Result Analysis Fig.2 Effect of sleep disorder and average work hours Fig.3 Chart of Job Pressure and Average work hours Fig.4. Effect of sleep disorder and Dependents Fig.5 Graph of Dependents and Job Pressure on Work performance

14. 04 Work-Perf Model Evaluation S/n Model Accuracy F-Measure Area under Curve (AUC) Root Mean Square Error (RMSE) Root Square Error Mean Square Error (MSE) Mean Average Error (MAE) 1. Support Vector Mech. 66.00 0.65 0.78 1.06 -0.45 1.12 0.84 2. K-NN 89.50 0.88 0.88 1.06 -0.47 1.14 0.86 3. Neural Network 80.80 0.80 0.82 1.12 -0.62 1.26 0.90 4. Ensemble-Bagged tree 89.10 0.89 1.00 1.72 -0.50 1.16 0.89 The model's accuracy was evaluated and compared to other machine learning models based on metrics as recorded in Table 2.

15. 04 Work- Perf’S Accuracy Comparison.

16. Findings and Conclusion 05

17. F indings and Conclusion. The absence of sleep disorder puts work performance rating at a higher rate than any other lesser value than the maximum value depicted in the graph. Contrast is the result of the predictive value of work performance generated when the individual average work hours are introduced as an associative factor value. T he effect of job pressure and average work hours on the predictive evaluation results as shown in Fig.3. The higher the assigned job pressures on workers or the volume of tasks, the lower their corresponding work performances. An individual with a very good sleep pattern, void of any form of disorder is potentially considered to be more productive than another with a sleep disorder. The possession of quite a large number of dependents by an employee of an organization has a negative influence on work performance as more effort will be directed towards the dependents affecting the individual's work performance. Conclusively, WORK-PERF an innovative artificial intelligent model for the automatic prediction of employee work performance ratings based on non-physical factors has been introduced. The evaluation results generated by the developed model are also significantly high, data collected for the developed model's training and testing are very comprehensive, and detailed with sampled instances of individuals from varied work spheres and different regions of the world. With this assertion, WORK-PERF becomes fit, applicable, and useful for the prediction of work performance in any region of the world. 05

18. References . Anna-Lina Müller: A tale of two genders: How women and men differ in their social policy responses to automation risk, Working Paper Series, 1-83, (2021). Ariane Hegewisch, Chandra Childers, & Heidi Hartmann: Women, automation, and the future of work - A report from institute for women’s policy research (2019). Benoit C, McCarthy B. & Jansson M.: Occupational stigma and mental health: Discrimination and depression among front-line service workers. Can Public Policy, 41: 61–69, (2015). Boudrias, J.-S., Desrumaux, P., Gaudreau, P., Nelson, K., Brunet, L., & Savoie , A.: Modeling the experience of psychological health at work: The role of personal resources, social-organizational resources, and job demands.  International Journal of Stress Management, 18 (4), 372 – 395, (2011).   https://doi.org/10.1037/a0025353 Chu, F., S. Gailus , L. Liu, & L.: The Future of Automated Ports . McKinsey & Company (2018). Falco, G., Viswanathan, A., Caldera, C., & Shrobe , H.: Master Attack Methodology for an AI-Based Automated Attack Planner for Smart Cities. IEEE Access, 6: 48360–48373, (2018). Ghafoor, Y., Huang, Y. P., & Liu, S. I.: An intelligent approach to discovering common symptoms among depressed patients. Soft. Comput ., 19(4), pp. 819 827, (2015). Gruetzemacher , R., Paradice , D. & Lee, K.B.: Forecasting extreme labor displacement: A survey of AI practitioners. Technological Forecasting and Social Change, 161-163, (2020) Hegewisch, A., Childers, C.& Hartmann, H: Women, Automation and the Future of Work. Institute for Women’s Policy Research (2019). Huo , W., W. Zhang, and P. S.-L. Chen.: Recent Development of Chinese Port Cooperation Strategies, Research in Transportation Business and Management 26: 67–75, (2018). Irene Y. Chen, Peter Szolovia & Marzyeh Ghassemi : Can AI Help Reduce Disparities in General Medical and Mental Health Care? AMA Journal of Ethics, 21(2): 167-179, (2019). Jean- Se´bastien Boudrias, Pascale Desrumaux, Patrick Gaudreau, Katia Nelson, Luc Brunet &Andre´ Savoie : Modeling the Experience of Psychological Health at Work: The Role of Personal Resources, Social-Organizational Resources, and Job Demands, International Journal of Stress, Vol. 18, No. 4, pp. 372–395, (2011). DOI: 10.1037/a0025353 Jeffrey Heer : Agency plus automation: Designing artificial intelligence into interactive systems. Proceedings of The National Academy of Sciences, vol. 116, 1844–1850, (2019).

19. Thanks Do you have any questions [email protected] [phone] ?