python - 增量分析中的特征缩放

Question

我正在对我的数据进行增量分析。数据属于 4 个年龄组（第 1 天、第 2 天、第 3 天和第 4 天）。在将数据输入模型之前，我使用 sklearn 中的标准缩放器实现对特征进行标准化。当我想到它时，我想到了三种方法。

Approach (1)standardize the newly added data separately
days = [day1, day2, day3, day4]

data=[]
for day in days:
    standard_scaler = StandardScaler()
    scaled = standard_scaler.fit_transform(day)
    data.append(scaled)
    Y = model.fit_transform(data)

Approach (2)standardize all the data up to the current day together separately
days = [day1, day2, day3, day4]

data=[]
for day in days:
    data.append(day)
    standard_scaler = StandardScaler()
    scaled = standard_scaler.fit_transform(data)
    Y = model.fit_transform(scaled)

Approach (3)partial_fit the same standard scaler on the newly added increments
    days = [day1, day2, day3, day4]
    standard_scaler = StandardScaler()

    data=[]
    for day in days:
        standard_scaler.partial_fit(day)
        data.append(day)
        scaled = standard_scaler.transform(data)
       
        Y = model.fit_transform(scaled)

请告知哪种方法最适合。

Answer 1

方法 1 是最好的方法，实际上也是唯一正确的方法