Scholarlink

This research investigates the effectiveness of predicting future salaries for graduates using academic performance data from Sakon Nakhon Rajabhat University. The study applies five machine learning models-Decision Tree Regression, Random Forest Regression, Support Vector Regression (SVR), K-Nearest Regression, and Gradient Boosting Regression-on datasets both with and without dimensionality reduction. Principal Component Analysis (PCA) and Feature Selection were used as dimensionality reduction methods to streamline the dataset. However, the results indicate that dimensionality reduction did not improve prediction accuracy; in fact, the original dataset without dimensionality reduction yielded the lowest Root Mean Square Error (RMSE) across all models, with GBR achieving the best result at 1856.59. This finding suggests that dimensionality reduction may lead to a loss of crucial information, especially in datasets containing diverse academic records and course enrollment requirements. Consequently, future research should explore grouping similar data to increase predictive accuracy and applying multi-prediction models to each group. This approach may help retain important features and improve the effectiveness of salary predictions for graduate students.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Evaluating Salary Prediction Models for Graduates Using Dimensionality Reduction and Machine Learning Techniques

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Surasit Uypatchawong,
  • Kritanat Chungnoy,
  • Pokpong Songmuang

摘要

This research investigates the effectiveness of predicting future salaries for graduates using academic performance data from Sakon Nakhon Rajabhat University. The study applies five machine learning models-Decision Tree Regression, Random Forest Regression, Support Vector Regression (SVR), K-Nearest Regression, and Gradient Boosting Regression-on datasets both with and without dimensionality reduction. Principal Component Analysis (PCA) and Feature Selection were used as dimensionality reduction methods to streamline the dataset. However, the results indicate that dimensionality reduction did not improve prediction accuracy; in fact, the original dataset without dimensionality reduction yielded the lowest Root Mean Square Error (RMSE) across all models, with GBR achieving the best result at 1856.59. This finding suggests that dimensionality reduction may lead to a loss of crucial information, especially in datasets containing diverse academic records and course enrollment requirements. Consequently, future research should explore grouping similar data to increase predictive accuracy and applying multi-prediction models to each group. This approach may help retain important features and improve the effectiveness of salary predictions for graduate students.