- Have you ever noticed the normalization transformation but not understood what it does?
- Have you been normalizing in Spotfire with custom expressions?
This is week 3 of the Transformations series, and this week, I will discuss the normalization transformation. In case you are unfamiliar with what normalization is or what it is used for, normalization refers to adjusting a value. In most cases, this adjustment is to get from different scales onto a common scale. The use case that many oil and gas users will be familiar with is the need to normalize the start of production to a common date, so that you can see production for many wells starting at the same time. Normalization is also used as a data preparation task prior to running many predictive model. Predictive models often require data normalization because they are sensitive to outliers. If data was created on dramatically different scales, values could be interpreted as outliers rather than “regular” values, which would skew the results. Thus — NORMALIZATION!
Now, data can be normalized with many different methods. Spotfire offers 14 different methods as shown below.
Steps to normalize columns in a data set
- Go to the Insert menu, Transformations, and select Normalization from the drop down.
- Select either Add columns or Replace columns in the radio button setting depending on whether or not you want to retain the original data.
- Move the columns to be normalized from left to right.
- Choose the method from the drop down
- Choose how the columns should be named.
If you have chosen to create new columns, they will appear at the end of the data table. The Help menu has the detailed calculations of each of the 14 methods under “Normalizing Columns” as shown below.
Guest Spotfire blogger residing in Whitefish, MT. Working for SM Energy’s Advanced Analytics and Emerging Technology team!