You generally standardize in a multivariate analysis when you want all variables to be in comparable units. Normalize can mean different things. sometimes it means to fit a normal distribution to a set of data or also to transform a variable to a normal variable. Also in some disciplines normalize means to divide a sum or integral by a value that makes the resulting sum or integral one. HTH David

I. Normalization: data should be normalized if there is any predefined assumption of the model which you are going to use else there is no need. ANN doesn't have any assumption about the data.

II. Yes I agree with David, you need to standardized your data only when you are dealing with multiple variables together (i.e. Multivarite Anaysis). this is also valid for some other models not for ANN.

III. you need to transform that nominal (categorical) variable into numeric one, normal method in this case is to use 0-1 binary values for all the k-th category. let say your nominal variable have seven categories(days) then you have to take 7 dummy variable where each of them will take value 0 or 1 depending on their presence or absence.

The term normalization is used in many contexts, with distinct, but related, meanings. Basically, normalizing means transforming so as to render normal. Whendata are seen as vectors, normalizing means transforming the vector so that it has unit norm.

Part of the derivation process, standardization is the process by which similar datareceived in various formats is transformed to a common format that enhances the comparison process. For example, street names commonly contain directions, like North or West

Prity Kumari : there is an assumption on standardization that it should be normally distributed. Hence, should we apply standardization to dataset that is non-normally distributed? I want to integrate 2 data sources (different scales) to feed ANN, should I use normalization or standardization?

Thanks

Normalizing the data is sensitive to outliers, so if there are outliers in the data set it is a bad practice. Standardization creates a new data not bounded (unlike normalization).

for NN the Standardization is better

It is recommended to adopt standardization, reason being the models built on standardized data give better results(predictions).

Use normalization when the features data can not be normalized or impossible to make them follow Gaussian distribution.

Standardization of data is when you need to do multivariate analysis on data of different units, while if your data is not normally distributed you need to do normalization before running statistical analysis. So they are different processes you can do both on your data before you apply statistical analysis.