Yasaman Malekiyourtchi

I hold a Ph.D. in Statistics from the Department of Statistics at Amirkabir University of Technology (Tehran Polytechnic) and have more than ten years of academic, teaching, and research experience as an Assistant Professor of Statistics at Alzahra University, Tehran, Iran.

My Ph.D. research focused on locally self-similar processes, a subclass of non-stationary stochastic processes, with particular emphasis on time-frequency spectrum estimation. These processes are relevant to real-world applications such as financial markets, internet traffic, signal processing, and other dynamic systems in which statistical properties change over time. This foundation enables me to understand the mathematical structure, assumptions, limitations, and uncertainty of statistical and machine-learning models and to apply them appropriately to different scientific problems.

My publication record includes peer-reviewed journal articles in probability and statistics, as well as conference proceedings. In addition to my doctoral research, I have worked on probability density function estimation for non-stationary processes. This is particularly relevant to signal processing and streaming-data applications, such as internet traffic and transaction data, where probability distributions and dependence structures can evolve over time.

I have extensive university teaching experience across a broad range of theoretical, computational, and applied statistics courses. These include Probability and Measure Theory at graduate level, Probability Theory, Stochastic Processes, Time Series Analysis, Mathematical Statistics, Statistical Methods, Computational Statistics with R, Design and Analysis of Experiments, and applied statistics for students in engineering, economics, psychology, and other disciplines.

I have supervised both graduate and undergraduate students in their research projects, supporting them in defining research questions, selecting suitable statistical methods, analysing data, interpreting results, and preparing their theses and reports. I also served as Head of the Department of Statistics for four years.

Research and Professional Activities

Statistical Modelling in Fisheries Research
* Worked with fisheries stock-assessment models, including the State-space Assessment Model (SAM) and Stock Synthesis (SS3), using R.
* Contributed statistical expertise to research on cod, saithe, shrimp, and pollack stocks.
* Applied time-series models, regression models, generalized linear mixed models, generalized additive models, spatio-temporal methods, and probabilistic approaches to fisheries and survey data.
* Worked with biological and acoustic data for stock analysis, prediction, and abundance-index estimation.
* Assessed model assumptions, parameter sensitivity, uncertainty, model diagnostics, and predictive performance.
* Developed reproducible workflows in R for data preparation, modelling, validation, prediction, visualisation, and scientific interpretation.

Machine Learning and Oceanographic Data Analysis
* Work as a member of the AI and Analysis Group, applying my background in statistics, probability theory, stochastic processes, and time-series analysis to machine learning and advanced oceanographic data analysis.
* Analyse large and complex oceanographic datasets containing spatial, temporal, environmental, and vertical information.
* Contribute to the selection, training, tuning, validation, and interpretation of machine-learning models for automated quality control, anomaly detection, and downscaling of ocean-model outputs.
* Work with machine-learning approaches including regularised models, tree-based methods, XGBoost, and methods for high-resolution reconstruction of oceanographic fields.
* Incorporate scientific and physical knowledge into feature construction and model evaluation so that predictions are not only accurate but also meaningful from an oceanographic perspective.
* Apply explainable-AI methods to understand model behaviour, identify influential variables, and evaluate the scientific plausibility of predictions.
* Address important modelling challenges such as extreme class imbalance, missing data, spatial and temporal dependence, information leakage, uncertainty, overfitting, and generalisation to new data.
* Design validation strategies and performance measures according to the structure of the data, the scientific question, and the intended use of the model.
* Develop approaches for monitoring data drift and changes in model behaviour when new expert-validated labels are delayed or unavailable.
* Combine statistical theory with machine-learning methods rather than applying AI models as black-box tools.

Core Competencies
* Probability theory and stochastic processes
* Time-series analysis and forecasting
* Statistical inference and modelling
* Non-stationary processes and time-frequency analysis
* Regression and mixed-effects models
* Generalized additive and spatio-temporal models
* Probabilistic modelling and uncertainty assessment
* Fisheries stock-assessment models, including SAM and SS3
* Fisheries, acoustic, and oceanographic data analysis
* Machine-learning model selection, training, tuning, and validation
* Anomaly detection, model monitoring, and data-drift assessment
* R programming and reproducible data analysis
* University teaching and student supervision