Creating features with general and cumulative operations
The featuretools library uses what are called transform primitives to create features. Transform primitives take one or more columns in a dataset as input and return one or more columns as output. They are applied to a single DataFrame.
The featuretools library divides its transform primitives into various categories depending on the type of operation they perform or the type of variable they modify. For example, general transform primitives apply mathematical operations, such as the square root, the sine, and the cosine. Cumulative transform primitives create new features by comparing a row’s value to the previous row’s value. For example, the cumulative sum, cumulative mean, and cumulative minimum and maximum values belong to this category, as well as the difference between row values. There is another cumulative transformation that can be applied to datetime variables, which is the time since previous transformation...
