Optimizing Personality Prediction from Resumes Using Novel Random Forest Algorithms in Comparison with K Nearest Neighbor Algorithm to Improve Accuracy
K. Maheswar Reddy, Ramachandran Thandaiah Prabu, A. Ezhil Grace · 2023
The aim of this paper to assess and enhance prediction accuracy in determining a candidate's personality from their CV by comparing the efficacy of the Random Forest algorithm and the K Nearest Neighbour algorithm. In total, 80 samples were gathered and segregated into two groups, each comprising 40 samples. The second group utilized the K Nearest Neighbour Algorithm, while the first group employed a random forest model. The dataset was integrated into the study using the Kaggle tool and trained using the most advanced random forest technique within a Jupiter Notebook environment. The determination of the sample size was grounded in prior research for a pre-test power of 95% and an alpha value of 0.604. The simulation results conclusively surpassed the K Nearest Neighbour technique, attaining accuracy scores of 90.9970% and 87.8140%, respectively. This difference in accuracy between the two methodologies with a value of 0.013 (P<0.05). Hence, within the given dataset context, machine learning approaches such as Random Forest and K Nearest Neighbour exhibit promise in accurately estimating personality traits based on CVs. The findings emphasize the potential of these algorithms in advancing the precision of personality prediction from CV data.