體重預測模型比較

Weight Prediction Model Comparison

[Collecting data]

這個模型建立在每天早晚體重差、以及過去趨勢之上。只測量單日的變化，能貼近真實的生活習慣。實驗的前十二週，只會用統計模型預測，同時持續累積資料；等資料量夠了，才會開始訓練機器學習模型。並開始比較兩個模型的差異。

This model is built on the morning/evening weight difference each day, plus the trend over time. Measuring only the single-day change actually makes the model fit my real habits and body rhythm more closely. Statistical and machine learning models are two completely different approaches, so for the first twelve weeks of the experiment I'll only use the statistical model while data keeps accumulating; once there's enough data, I'll start trying the ML model, and the comparison between the two will begin after that.