94280 - Data Management

Learning outcomes

Students will be aware of the importance to manage data sets for an efficient chemical assessment. They are expected to be able to: 1. Use statistical methods to validate analytical data and to in findi out correlation and trends; 2. Identify the best experimental framework for an efficient data collection; 3. Use chemometrics techniques to extarct information from large datasets.

Course contents

The CU is composed of three modules with the following contents.

Measuring Variability and Statistical Decision.

1.Deviations in experimental results: random, systematic and gross.
2. How to measure and how to minimize random and systematic deviations. Parameters for measuring data dispersion. Parameters for measuring data location.
3. The concept of probability and the role of statistics in Quality Control.
4. Random distributions. Gaussian distribution. The Central Limit Theorem.
5. Confidence intervals.
6. Statistical decision tests: comparison of a mean to a reference value or comparison of two means (z-tests and t-tests, including paired t-test), comparison of variances (F-test), comparison of distributions (chi-square tests) and tests for outliers (Q-test).
7. Analysis of variance: use of one-way and two-ways ANOVA in the analysis of data variability caused by different sources.

Experimental Design

1. Description of a general optimization flowchart.
2. Assessment of optimization criteria. Single and multi-objective responses.
3. Application of screening methods for searching for relevant variables.
4. Application of statistic methods focused on the study of factors and interactions.
5. Implementation of strategies for simultaneous optimization of interacting variables.

Chemometrics

1. Multivariate exploratory analysis: Principal Component Analysis. The concept. Interpretation: information about samples (scores), variables (loadings) and relationships among them. Preprocessing and outlier detection. Application to environmental data and to industrial process control.
2. Multivariate calibration: Partial Least Squares regression. The concept. Model complexity and quality parameters. Qualitative interpretation of the calibration model. Application to environmental data and to industrial process control.

Readings/Bibliography

Lecturer notes and slides

Teaching methods

The course unit is divided into three modules taught independently at different times in the academic year, each module is organized in theoretical classes where main concepts are introduced, as well as tutorial classes with discussion of case-study examples.

Assessment methods

Each module learning is evaluated independently, exploiting: i) written tests; ii) oral presentations or interviews; iii) written assignments or combinations of them. The Course Unit grade will be the arithmetic mean of grades from the three modules. ChIRS grades scale goes from 1 to 100, pass grade is
40, and will be translated into ECTS and different University scales. Criteria: knowledge on a very limited number of topics covered in the course and analytical ability that emerges only with the help of the instructor, using generally correct language → 40-45;
Knowledge on a limited number of topics covered in the course and independent analytical ability only on purely executive
issues, using correct language → 45-60;
Knowledge on a large number of topics covered in the course, ability to make independent critical analysis choices, mastery
of specific terminology → 60-80;
Essentially comprehensive knowledge on the topics covered in the course, ability to make independent critical analysis and
connection choices, full mastery of specific terminology, and ability for argumentation and self-reflection → 80-100.

Teaching tools

Lectures slides and notes will be available on the course moodle https://emmcchir-learning.ualg.pt

Office hours

See the website of Isabel Maria Palma Antunes Cavaco

See the website of Javier Vicente Saurina Purroy

See the website of Ana Maria De Juan Capdevila