MinMaxScaling.md (843B)
1 # Min-max scaling 2 3 ML CH2 4 5 **Definition:** Min-max scaling also referred to as normalization is a shift from the current values to between two arbitrary values. 6 7 These two bounds are normally either 0 and 1 or -1 and 1. It is optimal for neural networks to have zero mean inputs so a range from -1 to 1 is generally good. 8 9 This is often done by subtracting the min value and then dividing by the difference between the min and the max. 10 11 Here is an example implementation: 12 13 ```python 14 15 # For each column (assuming they are numbers) iterate through them and set all 16 # features to be equal to the (current - min) / diff. 17 # This has a lower bound of -1 and upper bound of 1. 18 19 for i in df: 20 min = df[i].min() 21 diff = df[i].max() - min 22 df[i] = (df[i] - min) / diff 23 24 df.describe() 25 ``` 26 27 See [FeatureScaling](FeatureScaling.md) for more.