AI in Scientific Research Course – AI and Data Research Masterclass: methodology, experiments, and machine learning

Training process

Training needs analysis

If you have specific requirements regarding the training programme, we will carry out a training needs analysis for you. This will guide us on which aspects of the programme should receive greater emphasis, so that the training programme meets your specific needs.

What will you gain?

Stronger study design - You will learn how to define goals, hypotheses and variables with precision, so you can design studies that are methodologically sound and easier to defend before peers or reviewers.

Effective questionnaires - You will create surveys and questionnaires with clear logic, suitable scales and solid validation, so the data you collect is consistent, interpretable and ready for meaningful analysis.

Coding qualitative data - You will turn statements and observations into structured categories and quantitative variables, allowing you to combine qualitative insight with basic statistics and clearer reporting.

Well-designed experiments - You will learn to plan A/B tests, RCTs and multifactor experiments, helping you choose variables correctly, avoid design flaws and interpret effects and interactions with confidence.

Large-scale data handling - You will master profiling, cleaning, merging and aggregating large datasets, so you can work with tables containing millions of records without losing structure, speed or data quality.

Stronger data analysis - You will run EDA, detect missing values, outliers and both linear and nonlinear relationships, so your modeling decisions are based on evidence from data rather than intuition alone.

Building ML models - You will practice feature preparation, variable selection and model training with decision trees, random forests and MLPs, giving you a practical pipeline for real research analysis tasks.

Reliable model evaluation - You will use metrics, cross-validation, hyperparameter tuning and ensemble methods, so you can compare models properly, reduce overfitting and choose solutions that remain robust.

Training programme

1. Introduction to research methodology

types of research: qualitative, quantitative, mixed,
principles of correct formulation of research objectives,
operationalization of variables and research hypotheses,
sample selection: probabilistic and non-probabilistic selection methods,
planning the schedule of the research project.

2. Creating questionnaires and surveys

structure of a typical questionnaire,
metadata and demographic section – selection of variables, design errors,
measurement scales:
- Likert scale (5-, 7-, 11-point),
- semantic, ordinal, dichotomous scales,
types of questions:
- closed-ended single-choice and multiple-choice questions,
- open-ended questions – rules of interpretation and coding,
- matrix and ranking questions,
branching logic, answer rotation, form validation,
questionnaire pilot test.

3. Quantitative interpretation of qualitative research results

coding techniques: open, axial, selective,
creation of analytical categories and subcategories,
semantic analysis:
- identification of themes and patterns of statements,
- concept maps and correlation networks,
operationalization of qualitative categories into quantitative variables,
basic statistical indicators on coded data.

4. Planning statistical experiments

types of experiments: A/B, quasi-experiments, field experiments,
defining hypotheses and selecting independent/dependent variables,
experimental designs: RCT, DOE (Design of Experiments), randomization,
designing multifactor experiments,
advanced analysis of results:
- ANOVA/ANCOVA,
- regression models and their interpretation,
- main effects and interactions,
- power analysis.

5. Working with large data sets

organization of work on data of the following size:
- > 75 columns,
- > 1 million tuples.
data profiling, anomaly detection, standardization,
memory optimization and work on large sets (chunking),
methods of joining, aggregation and filtering of large-scale data,
preparation of data for further modeling.

6. Machine Learning

exploratory data analysis (EDA):
- analysis of variable distributions (histograms, KDE),
- detection of missing data and methods of their imputation,
- correlations: Pearson, Spearman, VIF, correlation matrices,
- detection of nonlinear relationships,
data preparation and Feature Engineering:
- handling missing values, replacing outliers,
- normalization and standardization,
- encoding categorical variables: one-hot, target encoding,
- creating new features (feature engineering),
feature selection:
- filter methods: ANOVA F-score, chi-square,
- wrapper methods: RFE, boruta,
- embedded methods: Lasso, Ridge, L1/L2 regularization,
- dimensionality reduction: PCA, ICA, UMAP.,
building and training models:
- implementation and comparison of models:
  - decision tree,
  - random forest,
  - neural network (MLP),
- data split: train/test/validation,
- overfitting control and regularization,
evaluation and improvement of models:
- metrics: RMSE, MAE, R², accuracy, precision, recall (depending on the type of model),
- hyperparameter tuning: Grid Search, Random Search, Bayesian Optimization,
- ensemble learning: bagging, boosting (XGBoost, LightGBM), stacking,
- cross-validation and model stability.

What are the prerequisites for participating in the training?

Basic statistics - You should be comfortable with mean, median, standard deviation, correlation and basic charts, so you can follow result analysis and model evaluation without difficulty.

Tabular data handling - You should know how to work with data tables, filter records, recognize variable types and read CSV or XLSX files, so you can move smoothly into the analytical parts of the course.

Research logic - You should understand the difference between a research question, a hypothesis, a dependent variable and an independent variable, so you can follow study and experiment design.

Basic Python skills - You should know the basics of Python or a similar analytical environment, including running code and using libraries, so you can complete the modeling exercises with confidence.