Your friend's Mini$-$Project group selected a heart disease prediction dataset taken from a Kaggle repository for which a models had been implemented for a binary classification learning task. The two classes were "disease" $(95\%$ of the data$)$ and "healthy" $(5\%$ of the data$).$ As one of their contributions for the project, the group performed an exploratory data analysis, which led them to delete several features and to engineer several new ones. With this new version of the data, the group developed a new model using a stratified train$/$validation$/$test split of $60\frac{\%}{20}\frac{\%}{20}\%.$ In their final project report, they evaluated their new model by comparing it with the performance reported by the Kaggle authors. In testing. your friend's group achieved an accuracy of $94\%,$ while the accuracies reported by the Kaggle authors were no higher than $92\%.$

From this information, can we conclude that the group's new model is better than the models developed by the Kaggle authors? State your opinion $($yes or no$),$ and give at least three important reasons that justify it$.$