66295 - Chemometrics

Course Unit Page

  • Teacher Dora Melucci

  • Credits 6

  • SSD CHIM/01

  • Language Italian

Academic Year 2018/2019

Learning outcomes

Students  learn basics about theory and practice of Chemometrics for Analytical Chemistry.

At the end of the course, students have the following expertise: design of experiments; data processing by multivariate analysis; use of modern software for the application of mathematical and statistical methods.

Students will be able to apply the new skills to real problems concerning applications and research.

Course contents

Prerequisites

Students attending this course must have a good education in the fundamentals of analytical chemistry and instrumental analytical techniques.

Target

The course aims to provide students with the ability to design a chemical-analytical methodology from sampling to data analysis, starting from the design of experiments to arrive at the correct processing of data and presentation of the final technical report.

With these objectives, the following mathematical and statistical knowledge are provided:

Elements of multivariate statistical analysis
Methods of exploration of multivariate data
Multivariate modeling methods: multivariate classification and regression
Design of experiments (DOE)

The student will acquire the computer skills for the application of chemometric methods learned.

Finally, the student will develop the specific competence of a Chemometrician: optimize an entire chemical analysis process.

Contents

DATA EXPLORATION.

Multivariate structure of data. Matrices: dimension, transposition, centering, covariance, correlation. Pretreatment of the data. Transformation of variables. Handling of missing data.

Principal component analysis. Loading plots. Score plots. Choice of principal components (rank analysis), both numerically and graphically (scree plot).

Clusters analysis. Distance matrix, similarity matrix. Agglomerative hierarchical methods for the analysis of clusters. Dendrograms.

MODELLING

Models. Order and linearity of a model. Control parameters. Validation of a model.

Classification: qualitative models. Confusion matrix. Loss matrix. Control parameters. Misclassification risk (MR%). Classification by K-NN. Discriminant analysis (DA). Classification by SIMCA. Classification by CART.

Calibration: quantitative models. Linear regression: MLR method. Leverages. Regression coefficients. Evaluation parameters for a regression model. Correlation coefficient. Prediction coefficient. Standard error of the estimate. Diagnostic methods for regression models. Principal Component Regression  (PCR). Partial Least Squares Method (PLS). Practical examples of calibration by means of PLS regression: spectrophotometry, pulsed stripping voltammetry, chromatography-mass spectrometry.

DESIGN OF EXPERIMENTS

Multivariate methods for the selection of standard samples and variables for model creation. Full Factorial Design. Fractional Factorila Design. Mixtures Design.

Readings/Bibliography

-       Roberto Todeschini, Introduzione alla chemiometria, Edises, 1998.

-       J.C. Miller, J.N. Miller, Statistics and Chemometrics for Analytical Chemistry, Pearson Education, 2010.

-       Richard G. Brereton, Applied Chemometrics for Scientists, Wiley, 2007.

-       Richard Kramer, Chemometric techniques for quantitative analysis, Marcel Dekker, 1998.

Teaching methods

The course consists of lectures (32 hours) and exercises in the computer lab (24 hours).

Lectures are dedicated to the acquisition of the basic concepts of Chemometrics and to the acquisition of specific informatic tools (software for mathematics and statistics).

Exercises in the computer lab are designed to enable students to use Chemometrics tools and to apply concepts and software to solve real problems of multivariate chemical analysis.

Crucial will be the use of material provided by the lecturer made ​​available online (http://campus.unibo.it/) and lecture notes.

Assessment methods

At the end of the course, students must deliver a report about a chemometric problem, relevant to a dataset provided by the teacher. Problems are individual: each student works on a different dataset. The chemometric data-processing in the final report is similar to what explained during the course, in guided exercises relevant to model-datasets. The report must be in the form of a text document. It is NOT required to deliver file corresponding to the numerical calculations, while it is required that numerical or graphical outputs be included . The teacher assigns a mark to the final report.
The examination consists of oral questions about the final report and the theory explained in the room lessons (definitions and demonstrations). The teacher assigns a mark to the oral examination.
The final mark is the average of the mark assigned to the report and the mark assigned to the oral exam.

Teaching tools

Blackboard for theoretical lessons. Video projector for explanation of spreadsheets. Informatic laboratory for exercises.
For lectures and exercises the teacher uses the following programs: Microsoft Excel and The Unscrambler.
To carry out individual exercises and calculations for the final report, students can use the PCs of the informatic laboratory. Outside the dates scheduled for the course, students can access the informatic lab in the presence of the teacher.

Links to further information

http://campus.unibo.it

Office hours

See the website of Dora Melucci