Predicting AI Service Focus in Companies Using Machine Learning: A Data Mining Approach with Random Forest and Support Vector Machine

Abstrak

This study investigates the prediction of AI service focus in companies using machine learning models. The primary objective is to predict the percentage of AI service focus based on company characteristics such as project size, hourly rate, number of employees, and geographical location. Two machine learning models, Random Forest Regressor and Support Vector Regressor (SVR), were trained and evaluated to determine their effectiveness in predicting AI adoption. The dataset consists of 3099 companies, with key features cleaned and preprocessed, including the transformation of categorical variables into numerical ones using one-hot encoding and imputation techniques applied to handle missing values. The Random Forest model demonstrated better performance, with an R² value of 0.12, indicating a modest ability to explain the variance in AI service focus. In contrast, the SVR model had a negative R² value of -0.03, suggesting that it struggled to capture the underlying relationships in the data. The analysis identified project size and hourly rate as the most significant predictors of AI service focus, with larger projects and higher hourly rates correlating with a greater emphasis on AI services. Despite the relatively low performance of both models, this research provides valuable insights into the factors that influence AI adoption. The findings emphasize the importance of project-related characteristics in determining a company's AI service focus. However, the study is limited by missing data and the absence of additional features that could further improve prediction accuracy. Future research could benefit from incorporating more business-specific features and advanced modeling techniques to enhance the predictive power and generalizability of the model.