I hope you are doing great, this blog is basically a continuity of previous blog "" and in this blog as I promised we will talk about normalization in detail and we will see what are the various types of normalization and when to use which type. So without any further delay let's get started.

💡

Before directly moving on to what are the various types normalization let recall what is normalization and when to even use normalization.

What is normalization and when to use ?

Normalizatoin is basically a feature scaling technique which is used in those kind of scenarios where we are not aware about the distribution of data or when our algorithm doesn't make any assumption about underlying data distribution and after applying normalization on our data or on some particular feature the values get transformed into new values ranging between 0 to 1.

💡

Now since you are aware about what does normalization means let's talk about various types of normalization.

Types of normalization

If you would see the scikit documentation for normalization you would find that there are various types of normalization, but for around 95% of the time you would use only the 4 main types of normalization. The 4 types of normalization which we will discuss in today's blog are: Min max normliaztion, Mean normalization, Max absolute normalization and robust scaling.

Min-max normalization

Min-max normalization is a data normalization technique which used to scale numeric values to a specific range, typically between 0 and 1. The process involves subtracting the minimum value from each data point and dividing it by the difference between the maximum and minimum values. Mathematicall it is defined as

💡

Min-max normalization can be beneficial for algorithms that require input data to be within a specific range, such as neural networks or distance-based methods.

Mean normalization

Mean normalization is another data normalization technique that is used to scale numeric values. It involves subtracting the mean (average) of the dataset from each data point and then dividing it by the range of the data i.e maximum - minmum value.

Mean Normalization and Feature Scaling — A simple explanation | by K. Adith Narasimhan | Analytics Vidhya | Medium

💡

Mean normalization can be useful in scenarios where the data has a non-zero mean or exhibits a certain pattern or trend. It can help remove bias from the data and make it more comparable or suitable for certain statistical analyses or machine learning algorithms. Additionally, mean normalization can be beneficial in cases where the range of values is large, and min-max normalization may result in significant scaling differences.

Max absolute scaling

Maximum Absolute Scaling is a data scaling technique used to normalize numerical data by dividing each data point by the maximum absolute value in the dataset. It scales the data to a range between -1 and 1. Mathematicall defined as

💡

Maximum Absolute Scaling is commonly used in scenarios where the scale of the data is important, and the sign or the magnitude of the values needs to be preserved, such as in some machine learning algorithms or when dealing with specific features or measurements that have a known maximum absolute range.

Robust scaling

Robust Scaling, also known as Robust Standardization or Median and MAD Normalization, is a data scaling technique used to normalize numerical data while being robust to outliers. It is particularly useful when dealing with datasets that contain extreme values that can significantly impact the scaling process.

💡

It is particularly effective when the data contains outliers that may affect the scaling process and when the data does not follow a normal distribution.

Short Note

I hope you good understanding of what is normalization, what are the various type fo normalization and when to apply which normalization technique. So if you liked this blog or have any suggestion kindly like this blog or leave a comment below it would mean a to me.

💡

Also I would love to connect with you, so here is my Twitterand linkedin

Types of normalization in Machine learning

Table of contents