Predictive analysis of cross-cultural issues in global software development using AI techniques

The objective of this thesis is to identify predictive modelling that can be employed to mitigate problems due to cultural differences in GSD environment. When developing software for global usage, more and more teams consist of members that have different cultural backgrounds. Such differences cause communication barriers, have conflicting approaches to work, and generate confusion that results in weaker projects. The objective of this study is to arrive at risk factors useful in early detection of cross-cultural impacts so that important problems do not compromise project results. The current algorithms tested were Linear Regression, Ridge Regression, Lasso Regression, SVR, and XGboost. A set of criteria was adopted for the comparison of each model which includes Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE) and R-squared. Among all, Linear Regression returned the highest value of R-squared = 0.90 which shows that the model can predict with high accuracy. As for other models, medium accuracy was revealed by both Ridge Regression and XGBoost but adjusting them boosted the results only slightly. The analysis further shows that Linear Regression is the most accurate model in this task. The work offers important implications for the integration of existing AI-based methods into human-centered approaches that can support cross cultural concerns addressed in software development teams. When conflicts of interest are foreseen, approaches for improved understanding and cooperation should be employed to increase the efficiency of the team's work and the success of a project. Keywords: Cross-Cultural Challenges, Global Software Development, Predictive Analysis, Machine Learning, AI-Driven Solutions.