Is every process (such as data acquisition, splitting the data for validation, data cleaning, or feature engineering) that is done on the data before we train the model always called the pre-processing part? Or are there some processes that are not included?
Data preprocessing consists of all those techniques used to generate the final datasets (with an appropriate size, structure, and format) for the machine learning algorithms or models. Data acquisition should not be part of data preprocessing, but the step preceding it, which gathers the raw data (which may e.g. be noisy).
The book Data Preprocessing in Data Mining (2014), by Salvador García et al., which provides a good overview of the data preprocessing techniques and their connection with data mining and machine learning algorithms and models, defines data preprocessing as follows.
Data preprocessing includes data preparation, compounded by integration, cleaning, normalization and transformation of data; and data reduction tasks; such as feature selection, instance selection, discretization, etc. The result expected after a reliable chaining of data preprocessing tasks is a final dataset, which can be considered correct and useful for further data mining algorithms.
From page 10 onwards, there is a description and categorization of the main data preprocessing techniques. I will just list them, so refer to the book for a definition and explanation of each of these techniques.
- Data Preparation
- Data Cleaning
- Data Transformation
- Data Integration
- Data Normalization
- Missing Data Imputation
- Noise Identification
- Data Reduction
- Feature Selection
- Instance Selection
- Feature Extraction/Instance Generation
Here are two screenshots (from the cited book) that illustrate some of the data preparation
and data reduction techniques.