Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Altynshash Rakhimzhanova has a total of 7 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Short-Term Hydropower Generation Forecasting for Operational Planning and Early Energy Procurement: Multi-Model Evidence from Kazakhstan

    by Altynshash Rakhimzhanova, Nurkhat Zhakiyev, Aliya Nugumanova

    Reliable short-term hydropower forecasting is essential for dispatch planning and early electricity procurement in snowmelt-influenced power systems. This study develops a leak-free operational forecasting framework using quality-controlled hourly generation and hydro-meteorological records from eight hydropower plants in Kazakhstan. Two tasks are addressed: deterministic multi-step forecasting for D+1–D+7 and uncertainty-aware envelope forecasting for D+8–D+14 using MIN and Q90 targets. The benchmark uses Persistence as the primary baseline, against which RIDGE, SARIMAX, Random Forest, HistGradientBoosting, MLP, and LSTM are compared using Nash–Sutcliffe efficiency (NSE), root mean squared error (RMSE), and mean absolute error (MAE). For D+1–D+7, the results reveal strong cross-station heterogeneity and the expected decline in skill with increasing lead time. In the aggregated comparison, SARIMAX achieves the highest mean NSE at D+1 (0.903), while RIDGE becomes strongest by D+7 (0.625), both outperforming Persistence (0.534 at D+7). At the station level, SARIMAX performs best for Kapch, Kask, Moin, Bukh, and Ustk, RIDGE is best for Shar and Lenin, and LSTM is best for Shulb. The strongest stations, Kapch and Kask, reach mean NSE values of 0.941 and 0.933, respectively, whereas Ustk and Bukh remain the most difficult cases. A central methodological contribution is a flood-sensitive switched hybrid strategy for Ust-Kamenogorsk based on an observed-generation high-flow window selected by a regime-score procedure. This strategy improves robustness at medium lead times: for SARIMAX, NSE increases from 0.587 to 0.739 at D+2 and from 0.161 to 0.559 at D+7, while for RIDGE, NSE increases from 0.549 to 0.701 at D+2 and from 0.109 to 0.435 at D+7, together with substantial RMSE and MAE reductions. For D+8–D+14, envelope forecasting remains informative, but model ranking becomes target-dependent: SARIMAX and RIDGE provide the strongest mean performance for MIN (0.664 and 0.658), whereas LSTM and RIDGE are strongest for Q90 (0.746 and 0.743). Overall, the results show that hydropower forecasting in Kazakhstan is best approached as a station-wise, regime-aware, and horizon-specific problem.

  • Open Access

    PHYSICALLY BASED EVALUATION OF SNOWPACK SENSITIVITY TO TEMPERATURE PERTURBATIONS IN EAST KAZAKHSTAN

    by Altynshash Rakhimzhanova, Anar Rakhimzhanova, Bolat Kassyiet, Zheniskul Zhantassova, Kudaibergen Zhanpeissov

    Seasonal snowpack is a critical regulator of water supply and flood risk in continental climates, yet its reliable assessment in Central Asia is constrained by sparse observations. This study applies the multilayer Snow Thermal Model, driven by ERA5-Land reanalysis, to simulate snowpack evolution in East Kazakhstan during the 2022-2023 season and evaluates its performance against snow depth and snow water equivalent observations from the Kazhydromet network. The model reproduced snow accumulation, peak storage, and melt onset with high accuracy, achieving explained variance above 90%. Importantly, analysis of energy fluxes and stratigraphy revealed that more than half of simulated meltwater was produced under subfreezing air temperatures. Snowmelt is primarily controlled by positive surface energy balance dominated by net radiation and turbulent heat fluxes. Perturbation experiments further highlight the disproportionate sensitivity of the snow regime to modest thermal changes. A uniform +2 °C warming reduced peak snow water equivalent by nearly one third and advanced melt onset by two to three weeks, while a −1 °C cooling increased snow storage and prolonged snow duration. These threshold-driven responses show that even small climatic deviations or biases in forcing data can shift runoff timing and seasonal water availability. For water managers, this implies that operational planning must explicitly account for temperature sensitivity, since minor departures from average conditions can trigger substantial changes in spring flood risk. Overall, the study demonstrates that reanalysis-driven, physically based snow modeling provides robust diagnostics in data-scarce regions, surpassing empirical methods in both accuracy and explanatory power. The findings establish its importance for climate sensitivity analysis, flood preparedness, and water resource planning in snow-dominated basins.

  • Open Access

    COMPARATIVE STUDY OF PARALLEL ALGORITHMS FOR MACHINE LEARNING METHODS

    by Nurislam Kassymbek, Altynshash Rakhimzhanova

    In the modern world, as the amount of data used in machine learning is constantly growing, the task of accelerating the training of models on large datasets becomes relevant. To solve this problem, methods of parallel data processing are used. This paper discusses methods of parallel data processing for machine learning. Linear regression and random forest are considered as machine learning methods. Parallel algorithms based on the MPI interface were developed for each method. The results of the experiments showed that both methods give acceleration compared to the sequential algorithm. However, the acceleration in the case of random forest was significantly higher than in the case of linear regression. This is because random forest is a more computationally efficient method than linear regression. Therefore, it can be concluded that Random Forest is the most effective machine learning approach for parallel data processing. This statement is confirmed by the results of experiments conducted in this work. Overall, the experimental results show that the use of parallel algorithms in machine learning can significantly speed up model training when working with large data sets. Random forest is the most efficient method for parallel data processing, as it is more computationally efficient and has higher scalability.