When working on Machine Learning projects, we often need to adjust the data to make algorithms run faster and more efficiently. This adjustment involves methods like Normalization and Standardization.

Normalization

$$X normalized = X-Xmin/Xmax- Xmin$$

X = original value , Xmin = minimum value in the dataset , Xmax = maximum value in the dataset.

For example, imagine a dataset with two columns: Salary and Years of Experience. While "Experience" might range from 1 to 10 years, "Salary" could be in the thousands, like 10000 or 20000. These values are on very different scales, which can confuse the learning algorithm.

To solve this, we use Normalization (also called Min-Max Scaling). This method resizes the values to fit within a specific range, like [0, 1] or [-1, 1]. By bringing the data into a similar scale, the algorithm can focus better on patterns in the data, improving both speed and accuracy.

In short, when the data's range is uneven, we normalize it to make it easier for the algorithm to learn.

Standardization

$$Xstandardized = X−μ/σ ​$$

μ = mean , σ = standard deviation

Standardization is a data scaling technique that adjusts the values in a dataset so they have a mean (μ) of 0 and a standard deviation(σ )of 1. This process transforms the data into a standard normal distribution (bell curve shape). It is also called Z-score normalization.

Use it when data has varying scales for example, height in centimeters and weight in kilograms or standardization is preferred for a feature if it can sometimes have extremely high or low values.

Conclusion

There’s no strict rule for when to use these techniques; it depends on the situation. The explanations above are just basic guidelines. In practice, you can try both normalization and standardization and see which one works better for your specific problem.

But here are some points which you can also consider

• Many algorithms like logistic regression, support vector machines, and principal component analysis perform better with standardized data.

• Use when data has outliers because standardization is less sensitive to outliers compared to normalization.

• Unsupervised learning algorithms like Clustering, in practice, more often benefit from standardization than from normalization