AI impact on HR evaluation ppt for ADA.pptx
Download as PPTX, PDF0 likes18 views
The document discusses using machine learning approaches for predictive analytics in HR. It identifies gaps in traditional manual HR data analysis approaches and outlines objectives to apply machine learning algorithms to analyze employee information and classify employees. Unsupervised K-means clustering is used to partition employee salary data into clusters. A random forest classifier then determines which cluster an individual employee falls into based on their monthly salary. The study automates determining where HR needs additional hiring or promotions based on clustering attributes. It demonstrates classifying employees into salary-based clusters but could be expanded to consider additional performance variables.
More Related Content
Similar to AI impact on HR evaluation ppt for ADA.pptx (20)
Recently uploaded (20)
AI impact on HR evaluation ppt for ADA.pptx
- 1. Predictive analytics of HR A machine learning approach Published in - Journal of Statistics and Management Systems Author: V. Kakulapati , Kalluri Krishna Chaitanya , Kolli Vamsi Guru Chaitanya & Ponugoti Akshay Link:https://doi.org/10.1080/09720510.2 020.1799497 Published in : Jul 2020
- 2. Evolution- A step by step metamorphosis HR Analytics - new impetus Changing business scenario. Globalization and managing global resources across domains. huge volume of data. Outcome analysis in HR analytics is gaining momentum for civilizing employee productivity to develop organizational growth. Traditional Approach slow burner Data analysis is still manual without automation Data processing is slow and poor in volume & accuracy. Unable to process large amount data due to lack of infrastructure. HR analysis is explanatory in nature , cannot be used to forecast future possibilities. IT advancement & automation Silver lining Data must be used for predictive analysis to manage uncertainty and chaos Data accuracy and precision required in decision making. Priority based predictive analysis to predict outcome is required to make strategies or provision for the unknown.
- 3. Related study A dawn of a new era Current literature review has revealed there is constant need felt for advanced data analytical techniques in HR operations. Large volume of HR estimates needs to be updated hence ML is used. These analytics assists companies containing HR-related expenses while improving business functioning in addition to employee commitment and satisfaction Predictive analysis in HR assist in executing the HR functions like predictive market trends, employee salary benchmarking, recruitment and forecasting. Used for enhancing performance of employees and organizational policy change to meet uncertainty and chaos. It alters and increase innovation and can reach an accuracy of almost 100 percent in decision making and resource allocations. Enables management with estimations of multiple variables that would impact business and draw conclusive reports based on the analysis.
- 4. Gap analysis and Objectives Gaps identified by the author:- The assortment, processing and data analysis mostly in HR have been manual with limited variables. Aim of HR estimates is to assimilate paraments for enhancing productivity and performance, however parameters used are extremely generic and limited. In current practice various models have been built that merges analytical workforce data with other attitudinal variables to evaluate performance ,however these variable are not precise. Hence need of an investigation based competency model that accurately predicts employee performance. More studies required to use predictive analysis to access competencies & correlate them with performance factors. Few organization use performance data, employee past and present activities for career improvement, however they are not accurate and do not count for all aspects. Objectives as identified by the author:- Apply ML algorithm to analyze employee information for improving his/her position in the organization. To use predictive analysis in categorizing and classifying employees based on compensation, job performance , characteristics to payroll and service history to improve his growth prospects in the organization.
- 5. Variables and Data Modeling Variables Monthly Salary details/ compensation details Employee personal data as shown in the table. Organization attrition data Performance details of employee Data 50,000 records of which 1000 records are taken for investigation. 70% of data is considered as train data and remaining as test data. The data set is pre-processing for removing unnecessary elements or noise elements removal.
- 6. Methodology & Result Unsupervised clustering techniques K- means is used as a data mining technique. Data is partioned into K clusters with every item has a place in the cluster with the closest mean i.e k means. Data is clustered on the basis of monthly salary range of the organization. Using Random forest classifier based on monthly salary of employees, the relative position of the employee in the cluster is determined i.e in which cluster an individual employee falls. For e.g if an employee has Monthly Income approx. < 5000 falls under 1st cluster, between 7000 to 12000 fall in 2nd cluster and above 12000 in 3rd cluster.
- 7. What is Random Forest classifier & K clustering Random forest builds multiple decision trees and merges them together to get a more accurate and stable prediction. It is use for classification
- 8. Results and Conclusion The authors have used unsupervised learning algorithm to automate in which section & department HR needs to take additional employees. Using K clustering they have been able to define how many clusters there would be with the set attributes on which the HR is going to employ the new candidates or promote the existing candidates according to depending characteristics. The author have classified the clustering model based on Monthly income to determine under which cluster, which can be doing for any attribute depending on the feature set we are taking and the clustering.
- 9. Critical Evaluation The study misses out on some important performance variables like risk handling, decision making ability, managerial ability. A framework-based analysis of employee ability and performance analysis can be analysed can be further looked into. The literature review does not speak much of existing theories of ML in practice, neither does the authors quotes any such theory. Real time ML using HR practises in other organizations, does not find a mention in the article. The process of data interpretation and results is little vague and can be explained in details for better understanding.
- 10. Poushali Dey
Editor's Notes
- #7: Random forest classifiers : N estimator This is the number of trees you want to build before taking the maximum voting or averages of predictions. Higher number of trees give you better performance but makes your code slower OOB error rate - Out-of-bag (OOB) error, also called out-of-bag estimate, is a method of measuring the prediction error of random forests, boosted decision trees, and other machine learning models utilizing bootstrap aggregating (bagging)
- #8: Clustering is the process of dividing the entire data into groups (also known as clusters) based on the patterns in the data.