The Data Acquisition Pathway

We learn from data. Therefore, the way we prepare the data for the learning process will crucially condition the quality of learning and the reliability of the extracted knowledge. 

The first stage in data acquisition is the identification of the task that is, we have to know what kind of physical properties/biological activities we are going to model. 

Once we have defined this, we have to compile the initial dataset. First of aU, we decide upon the composition of the dataset. Usually, we take initially as many compounds as possible. 

The next and very important step is to make a decision about the descriptors we shall use to represent the molecular structures. In general, modeling means assignment of an abstract mathematical object to a real-world physical system and subsequent revelation of some relationship between the characteristics of the object on the one side, and the properties of the system on the other. 

Perhaps the best idea is to compute as many descriptors as possible and then to select an optimal subset by applying sophisticated techniques, discussed below. 

---

According to an elegant remark by Davies [5], "Modem scientific data handling is multitechnique, multisystem, and manufacturer-independent, with results being processed remotely from the measuring apparatus. Indeed, data exchange and storage are steps of the utmost importance in the data acquisition pathway. The simplest way to store data is to define some special format (i.e., collection of rules) of a flat file. Naturally, one cannot overestimate the importance of databases, which are the subject of Chapter 5 in this book. Below we discuss three simple, yet efficient, data formats. 

---