📄 中文摘要
在进行第一次机器学习项目时，模型在训练阶段显示出几乎完美的准确率，令我对结果充满信心。然而，当我在新数据上进行测试时，模型的表现却大相径庭。这让我意识到一个重要的教训：高准确率并不总意味着模型真正学到了东西。问题在于过拟合，由于数据集相对较小，模型开始记忆训练数据，而不是学习通用模式。它捕捉到了噪声、小的变化和不适用于新数据的特定细节，导致在新数据上的性能大幅下降。
当我的第一个机器学习模型记忆而非学习时（以及我如何解决这个问题）
📄 English Summary
When My First ML Model Memorized Instead of Learning (And How I Fixed It)
During my initial machine learning projects, the model exhibited nearly perfect accuracy during training, leading to a false sense of confidence. However, testing it on new data revealed significant issues. This experience highlighted a crucial lesson: high accuracy does not necessarily indicate that the model is genuinely learning. The core problem was overfitting. Due to a relatively small dataset, the model began to memorize the training data instead of identifying general patterns. It captured noise, minor variations, and specific details that were not applicable to new data, resulting in poor performance on unseen datasets.