ANALYSING AND PREDICTING THE IMPACT OF WORK FROM HOME TRENDS ON EMPLOYEE PRODUCTIVITY: A MACHINE LEARNING APPROACH

Abstract

The global COVID-19 outbreak has created lasting changes in how organizations operate, with remote work becoming a standard practice rather than an exception. Our research presents a computational model designed to anticipate worker output in home-based work scenarios. We analyzed data from 1,000 workers obtained from Kaggle's comprehensive repository, incorporating personal demographics, environmental workspace factors, and behavioral metrics. Our implementation utilized Linear Regression enhanced with preprocessing techniques—OneHotEncoder managed categorical data while StandardScaler normalized numerical inputs. Performance metrics revealed exceptional results: an R² value of 0.8434 indicates our model accounts for over 84% of output variation. Prediction accuracy measured through MAE (3.63 points) and RMSE (4.51 points) demonstrates reliability, with typical errors representing merely 3.6% deviation on our 100-point measurement scale. Through examining correlations and extracting feature significance, we identified critical factors driving remote work success. Our contribution advances current knowledge by transitioning from correlation-based observations to predictive forecasting, enabling evidence-driven policy formulation. Organizations can leverage these insights to enhance remote work strategies while maintaining worker satisfaction alongside operational efficiency.