Metrohm Vision Manual - Manual (Page 15)

▪▪▪▪▪▪▪ 13
2.2.3 Sample Selection
Sample selection is a necessary step in model development. In qualitative model development, all
spectra transferred to the library product form a temporary set and have to go through the sample
selection procedure. Sample selection divides them into a training set (good data), an acceptance set
(redundant samples), and a rejection set (outliers). Only when the sample selection results are saved,
can the model be developed from the training set. The same applies to quantitative model
development, where the analogous sets are called the calibration set, the validation set, and the
outlier set.
2.3 Data, Methods, and Results Storage
Understanding how Vision stores information will help in proper software operation.
Spectral data, constituent data, Data Collection Methods, and all parameters related to spectral
acquisition and manipulation (like dates and times of acquisition, audit trails) are kept in the Project
Database.
All qualitative methods (clustering, identification, and qualification method) and their parameters, as
well as the validation results are kept in the Library Database.
Calibration equations are kept in the Project Database.
Operations Methods are stored in the Project Database.
Results and spectra from Routine Analysis are stored in an output project database (created during
the Routine Analysis set-up) and/or in an ASCII file. ASCII files are not associated with any database
and can be stored in any location, including network drives.
2.4 User Preferences
The security database contains preferences for each registered Vision user. User preferences are a set
of default parameters applied every time a given user is logged into Vision. In general these
parameters include math pretreatments and methods used in the last method development session.
These preferences are updated every time the user edits a method and saves it. User preferences are
local, i.e., each part of Vision software may have its own set. Some programs do not support user
preferences and will start each time with the same defaults.
User preferences for qualitative and quantitative sample selection, qualification method development,
identification method development, and regression can be viewed and changed before method
development starts.
2.4.1 Intelligent default system
Intelligent default system facilitates fast new methods development and edition of existing methods.
Methods that are being redeveloped have old parameters as defaults. New methods take default
values from user preferences. When user preferences do not exist (a method of this kind has never
been developed in the past), the system defaults are applied.