Normalize/scale dataframe in a certain range-CodePudding

I have the following Dataframe:

pd.DataFrame({'DateTime': {0: Timestamp('2022-02-08 00:00:00'),
  1: Timestamp('2022-02-08 00:10:00'),
  2: Timestamp('2022-02-08 00:20:00'),
  3: Timestamp('2022-02-08 00:30:00'),
  4: Timestamp('2022-02-08 00:40:00')},
 'wind power [W]': {0: 83.9, 1: 57.2, 2: 58.2, 3: 48.0, 4: 69.5}})

             DateTime  wind power [W]
0 2022-02-08 00:00:00            83.9
1 2022-02-08 00:10:00            57.2
2 2022-02-08 00:20:00            58.2
3 2022-02-08 00:30:00            48.0
4 2022-02-08 00:40:00            69.5

As you can see, 83.9 is the maximum value in my second column and 48.0 the minimum value. I want to normalize these values in a range between 0.6 and 8.4, so that 83.9 would turn to 8.4 and 48.0 to 0.6. The rest of the numbers would fall somewhere in between. So far I only managed to normalize the column to a range of 0-1 with the code:

df['normalized'] = (df['wind power [W]']-df['wind power [W]'].min())/(df['wind power [W]'].max()-df['wind power [W]'].min())

I don't know how to further proceed to get these numbers in my desired range. Can someone help me, please?

CodePudding user response：

We can use MinMaxScaler to perform feature scaling, MinMaxScaler supports a parameter called feature_range which allows us to specify the desired range of the transformed data

from sklearn.preprocessing import MinMaxScaler

scaler = MinMaxScaler(feature_range=(0.6, 8.4))
df['normalized'] = scaler.fit_transform(df['wind power [W]'].values[:, None])

Alternatively if you don't want to use MinMaxScaler, here is a way scale data in pandas only:

w = df['wind power [W]'].agg(['min', 'max'])
norm = (df['wind power [W]'] - w['min']) / (w['max'] - w['min'])
df['normalized'] = norm * (8.4 - 0.6)   0.6

print(df)

             DateTime  wind power [W]  normalized
0 2022-02-08 00:00:00            83.9    8.400000
1 2022-02-08 00:10:00            57.2    2.598886
2 2022-02-08 00:20:00            58.2    2.816156
3 2022-02-08 00:30:00            48.0    0.600000
4 2022-02-08 00:40:00            69.5    5.271309

CodePudding user response：

You can use the wikipedia definition of feature scaling if you don't want to use sklearn:

a = 0.6
b = 8.4
x = df['wind power [W]']

df['normalized'] = a   (x - x.min()) * (b - a) / (x.max() - x.min())
print(df)

# Output
             DateTime  wind power [W]  normalized
0 2022-02-08 00:00:00            83.9    8.400000
1 2022-02-08 00:10:00            57.2    2.598886
2 2022-02-08 00:20:00            58.2    2.816156
3 2022-02-08 00:30:00            48.0    0.600000
4 2022-02-08 00:40:00            69.5    5.271309