Data scaling or normalization is a process of making model data in a standard format so that the training is improved, accurate, and faster. The method of scaling data in neural networks is similar to data normalization in any machine learning problem.

Some simple methods of data normalization are listed here:

• Z-score normalization: As anticipated in previous sections, the arithmetic mean and standard deviation of the given data are calculated first. The standardized score or Z-score is then calculated as follows:

Here, X is the value of the data element, μ is the mean, and σ is the standard deviation. The Z-score or standard score indicates how many standard deviations the data element is from the mean. Since mean and standard deviation are sensitive to outliers, this standardization is sensitive to outliers.

• Min-max normalization: This calculates the following for each data element:

Here, x_i is the data element, min(x) is the minimum of all data values, and max(x) is the maximum of all data values. This method transforms all the scores into a common range of [0, 1]. However, it suffers from outlier sensitivity.

• Median and MAD: The median and median absolute deviation (MAD) normalization calculates the normalized data value using the following formula:

Here, x_i represents each data value. This method is insensitive to outliers and the points in the extreme tails of the distribution, and therefore it is robust. However, this technique does not retain the input distribution and does not transform the scores into a common numerical range.