Forecasting Results Research Articles

When Infodemic Meets Epidemic: Systematic Literature Review.

Epidemics and outbreaks present arduous challenges, requiring both individual and communal efforts. The significant medical, emotional, and financial burden associated with epidemics creates feelings of distrust, fear, and loss of control, making vulnerable populations prone to exploitation and manipulation through misinformation, rumors, and conspiracies. The use of social media sites has increased in the last decade. As a result, significant amounts of public data can be leveraged for biosurveillance. Social media sites can also provide a platform to quickly and efficiently reach a sizable percentage of the population; therefore, they have a potential role in various aspects of epidemic mitigation. This systematic literature review aimed to provide a methodical overview of the integration of social media in 3 epidemic-related contexts: epidemic monitoring, misinformation detection, and the relationship with mental health. The aim is to understand how social media has been used efficiently in these contexts, and which gaps need further research efforts. Three research questions, related to epidemic monitoring, misinformation, and mental health, were conceptualized for this review. In the first PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) stage, 13,522 publications were collected from several digital libraries (PubMed, IEEE Xplore, ScienceDirect, SpringerLink, MDPI, ACM, and ACL) and gray literature sources (arXiv and ProQuest), spanning from 2010 to 2022. A total of 242 (1.79%) papers were selected for inclusion and were synthesized to identify themes, methods, epidemics studied, and social media sites used. Five main themes were identified in the literature, as follows: epidemic forecasting and surveillance, public opinion understanding, fake news identification and characterization, mental health assessment, and association of social media use with psychological outcomes. Social media data were found to be an efficient tool to gauge public response, monitor discourse, identify misleading and fake news, and estimate the mental health toll of epidemics. Findings uncovered a need for more robust applications of lessons learned from epidemic "postmortem documentation." A vast gap exists between retrospective analysis of epidemic management and result integration in prospective studies. Harnessing the full potential of social media in epidemic-related tasks requires streamlining the results of epidemic forecasting, public opinion understanding, and misinformation detection, all while keeping abreast of potential mental health implications. Proactive prevention has thus become vital for epidemic curtailment and containment.

Beyond the Road: A Regional Perspective on Traffic Congestion in Metro Atlanta

Traffic congestion not only affects traffic flow but also influences public perception of congested regions. While analyzing congestion at the road section level can help identify engineering solutions, it often fails to reveal broader spatial patterns and trends at the regional or macro scale unless summarized effectively. This study aims to address these challenges by focusing on regional-scale traffic congestion amounts measured by distanceTime metrics. A 12–month dataset, sampled every 10 min, was analyzed to identify spatial patterns, temporal trends, regional variations, and predictive models in the Metro Atlanta area. The results show that congestion is the most severe and increasing at key urban corridors like Brookhaven–Sandy Springs, the downtown connector, Druid Hills–Decatur, and Johns Creek–Cumming, aligning with recent urban developments. Cities such as Alpharetta, Dunwoody, Brookhaven, Austell, Stone Mountain, East Point, Lake City, Morrow, Fairburn, and Jonesboro show high increasing trends in congestion. Predictive modeling with the long short-term memory (LSTM) method shows promising results for short-term forecasts, though variability in data requires further optimization for certain cities. This research is significant because it demonstrates that congestion amounts measured by distanceTime metrics can be used for assessing regional characteristics broadly at a metropolitan city scale. The findings and methodologies identified in this research might support urban and transportation planning efforts in metropolitan planning organizations, such as the Atlanta Regional Commission, by identifying congestion amounts and trends at both the regional and road scales.

Predicting the popularity of music tracks on Spotify based on numerical metrics

In today's world, music plays an important role in the lives of millions of people, and music streaming platforms such as Spotify have become an integral part of modern culture. The popularity of music tracks is of great importance to the music industry, affecting artists' incomes and trends in the music world. Predicting the popularity of music tracks is an impor-tant task that can help artists, producers, and platforms better understand listener preferences and optimize their strategies. As part of this work, a data storage of music tracks on the Spotify platform has been de-veloped, based on a physical model of the database, the functionality of which is implemented using SQL scripts. Working with the database is presented through the implementation of software for the implementation of ETL processes and intelligent analysis of selected data. The software allows you to classify tracks by the level of popularity (0 - not at all popular, 1 - medium popularity, 2 - hit) using numerical track metrics such as acousticness, tempo, va-lence, liveness, etc. The role of the data storage management system is SQLite, the program-ming language for implementing the application is Python. Different machine learning models are used to predict track popularity, including KNeighbors, Decision Tree, Random Forest, and Extreme Gradient Boosting. Data mining software provides efficient track classification and graphical display, allowing users to easily interpret forecasting results. Libraries used in the work: pandas, numpy, seaborn, matplotlib, tabulate, xgboost, scipy, sqlite3. The overall analysis showed that the XGBoost and Random Forest models are the most effective for predicting the popularity of music tracks. They demonstrate high accuracy and resistance to changes in the set of attributes, which makes them suitable for use in real condi-tions.

Short-term load forecasting based on multi-frequency sequence feature analysis and multi-point modified FEDformer

Given the complexity and dynamic nature of short-term load sequence data, coupled with prevalent errors in traditional forecasting methods, this study introduces a novel approach for short-term load forecasting. The method integrates multi-frequency sequence feature analysis and multi-point correction using the FEDformer model. Initially, variational mode decomposition (VMD) technology decomposes the load sequence into multiple subsequences, each exhibiting distinct frequency characteristics. Subsequently, for each frequency band of the load sequence, the LightGBM algorithm quantifies the correlation between the load and various influencing factors. The filtered features are then input into the FEDformer model, providing preliminary short-term and long-term sequence prediction results. Finally, a point-by-point forecasting method based on a tree model generates multi-point load prediction results by training multiple LightGBM models. Throughout the forecasting process, a weighted threshold α is set, and a hybrid weighting method is utilized to combine the forecast results from different models, culminating in the final short-term load forecast results. Validation of the proposed hybrid model was conducted on an actual dataset from a specific area, The results exhibit higher prediction accuracy, affirming the proposed method as a novel and effective approach for short-term load forecasting.

Battery pack technological considerations for hybrid-electric regional aircraft feasibility

Abstract This paper presents a study of the effects of the durability and level of energy storage technology on energy management strategies and the performance of hybrid electric turboprops. The results highlight the key role of battery energy density on the durability of the battery pack and the viability of the concept of hybrid electric aircraft. Additionally, the trade-off between zero-day environmental compatibility and battery lifetime is identified, caused by the size of the pack. The effective energy density would decrease with the aging of the cells, leaving a significant inert mass and increasing fuel consumption. Optimal energy management strategies are suggested in light of this new information. Higher specific energy of the pack would mitigate this aspect, along with a reduction in fuel consumption and ${\rm{N}}{{\rm{O}}_{\rm{x}}}$ emissions. Indeed, the improvement of environmental compatibility was found to be nonlinear with a positive rate, suggesting high returns in investing in great improvements in energy density over a gradual increase. This result relates to the results of the statistical technological forecast presented in this study, which, without an increase in funding, predicts the availability of the specific energy required to match the fuel-only baseline in the 2040–2050 decade.

D-MGDCN-CLSTM: A Traffic Prediction Model Based on Multi-Graph Gated Convolution and Convolutional Long-Short-Term Memory.

Real-time and accurate traffic forecasting aids in traffic planning and design and helps to alleviate congestion. Addressing the negative impacts of partial data loss in traffic forecasting, and the challenge of simultaneously capturing short-term fluctuations and long-term trends, this paper presents a traffic forecasting model, D-MGDCN-CLSTM, based on Multi-Graph Gated Dilated Convolution and Conv-LSTM. The model uses the DTWN algorithm to fill in missing data. To better capture the dual characteristics of short-term fluctuations and long-term trends in traffic, the model employs the DWT for multi-scale decomposition to obtain approximation and detail coefficients. The Conv-LSTM processes the approximation coefficients to capture the long-term characteristics of the time series, while the multiple layers of the MGDCN process the detail coefficients to capture short-term fluctuations. The outputs of the two branches are then merged to produce the forecast results. The model is tested against 10 algorithms using the PeMSD7(M) and PeMSD7(L) datasets, improving MAE, RMSE, and ACC by an average of 1.38% and 13.89%, 1% and 1.24%, and 5.92% and 1%, respectively. Ablation experiments, parameter impact analysis, and visual analysis all demonstrate the superiority of our decompositional approach in handling the dual characteristics of traffic data.

Construction and Optimization of Integrated Yield Prediction Model Based on Phenotypic Characteristics of Rice Grown in Small–Scale Plantations

An intelligent prediction model for rice yield in small-scale cultivation areas can provide precise forecasting results for farmers, rice planting enterprises, and researchers, holding significant importance for agricultural industries and crop science research within small regions. Although machine learning can handle complex nonlinear problems to enhance prediction accuracy, further improvements in models are still needed to accurately predict rice yields in small areas facing complex planting environments, thereby enhancing model performance. This study employs four rice phenotypic traits, namely, panicle angle, panicle length, total branch length, and grain number, along with seven machine learning methods—multiple linear regression, support vector machine, MLP, random forest, GBR, XGBoost, and LightGBM—to construct a yield prediction model group. Subsequently, the top three models with the best performance in individual model predictions are integrated using voting and stacking ensemble methods to obtain the optimal integrated model. Finally, the impact of different rice phenotypic traits on the performance of the stacked ensemble model is explored. Experimental results indicate that the random forest model performs best after individual machine learning modeling, with RMSE, R2, and MAPE values of 0.2777, 0.9062, and 17.04%, respectively. After model integration, Stacking–3m demonstrates the best performance, with RMSE, R2, and MAPE values of 0.2483, 0.9250, and 6.90%, respectively. Compared to the performance after random forest modeling, the RMSE decreased by 10.58%, R2 increased by 1.88%, and MAPE decreased by 0.76%, indicating improved model performance after stacking ensemble. The Stacking–3m model, which demonstrated the best comprehensive evaluation metrics, was selected for model validation, and the validation results were satisfactory, with MAE, R2, and MAPE values of 8.3384, 0.9285, and 0.2689, respectively. The above research findings demonstrate that this integrated model possesses high practical value and fills a gap in precise yield prediction for small-scale rice cultivation in the Yunnan Plateau region.

Modelling of non-scheduled air transportation time series based on ARIMA

Forecasting non-scheduled air transportation demand is essential for effective resource allocation, operational planning, and decision-making. In this paper, the use of the ARIMA (Auto Regressive Integrated Moving Average) model for forecasting non-scheduled air transportation is explored. The ARIMA model is a widely employed time series forecasting technique which combines autoregressive (AR), differencing (I), and moving average (MA) components. It has been successfully applied to various fields and can be adapted to capture the patterns and trends in non-scheduled air transportation data. To forecast non-scheduled air transportation demand, historical data, including relevant variables are firstly collected. The data are processed by identifying and addressing any missing values, outliers, or trends that could affect the model's performance. Next, the ARIMA model is applied to the pre-processed data, utilising techniques such as model identification, parameter estimation, and model diagnostics. The ARIMA model captures the relationships between past observations and uses them to predict future demand for non-scheduled air transportation. The forecasting results from the ARIMA model provide insights into expected demand levels, peak periods, and potential fluctuations in non-scheduled air transportation. These forecasts enable decision-makers to optimise resource allocation, schedule aircraft availability, and enhance operational efficiency. However, it is important to note that the accuracy of ARIMA forecasts depends on various factors, including the quality and representativeness of the data, the appropriate selection of model parameters, and the stability of underlying patterns in the time series data. Regular model evaluation and refinement are crucial in maintaining forecasting accuracy.

Forecasting Half-Hourly Electricity Prices Using a Mixed-Frequency Structural VAR Framework

Electricity price forecasting has been a topic of significant interest since the deregulation of electricity markets worldwide. The New Zealand electricity market is run primarily on renewable fuels, and so weather metrics have a significant impact on electricity price and volatility. In this paper, we employ a mixed-frequency vector autoregression (MF-VAR) framework where we propose a VAR specification to the reverse unrestricted mixed-data sampling (RU-MIDAS) model, called RU-MIDAS-VAR, to provide point forecasts of half-hourly electricity prices using several weather variables and electricity demand. A key focus of this study is the use of variational Bayes as an estimation technique and its comparison with other well-known Bayesian estimation methods. We separate forecasts for peak and off-peak periods in a day since we are primarily concerned with forecasts for peak periods. Our forecasts, which include peak and off-peak data, show that weather variables and demand as regressors can replicate some key characteristics of electricity prices. We also find the MF-VAR and RU-MIDAS-VAR models achieve similar forecast results. Using the LASSO, adaptive LASSO, and random subspace regression as dimension-reduction and variable selection methods helps to improve forecasts where random subspace methods perform well for large parameter sets while the LASSO significantly improves our forecasting results in all scenarios.

Inventory optimization model using Artificial Neural Network method and Continuous Review (s,Q)

The medical device industry company experienced the problem of prolonged accumulation of finished goods in the warehouse, causing one of the safety box items to be defective and damaged. Therefore, this study aims to plan demand forecasting and design inventory policies that consider repair items caused during the buildup of finished goods in the warehouse to minimize total inventory costs using ANN and Continuous Review (s,Q) methods. Demand forecasting is carried out for the next 20 months, from May 2023 to December 2024, using the ANN model with a total forecasting of 17936 units of inner items and 3370 units of outer items. After that, the inventory policy calculation uses the continuous review (s,Q) method. The calculation results show a decrease in the total inventory cost on inner items by 83% and outer items by 79%. After demand forecasting, there was also a decrease in the total initial inventory cost of inner items by 81% and outer items by 80%. This research develops an inventory optimization model that considers repair items due to the accumulation of goods in the warehouse by integrating holding cost, ordering cost, and repair cost variables to develop inventory policies to be more effective and efficient and to utilize damaged products for repair and resale. The limitation of this research is that it only gets demand forecasting results for the next 20 months because the company only started operating in September 2021 and limited data access. It is hoped that future researchers can plan and design an inventory policy strategy with demand forecasting for the next 10 years, focusing on repair items caused by the accumulation of finished goods in the warehouse.

Cost Benefit Analysis of Energy Retrofitting with the Addition of Ventilation Holes in Middle Low-Income Houshold Buildings

The current climate change is characterized by an increase in the earth's surface temperature; one of the impacts that is felt quite significantly is the thermal comfort conditions in buildings. This will encourage residents to use air conditioning to provide the required level of thermal comfort. From the results of electricity demand forecasts, from 2017 to 2036, electricity demand will increase by 6.4% annually, dominated by the household sector at 38.49%. Efforts can be made to condition the room's thermal comfort to reduce the AC's operating time. This can be realized through building energy retrofitting, especially in middle-low-income households. Retrofitting will also not require high costs, so it suits middle-low-income households. In this research, retrofitting was carried out by implementing ventilation. It is hoped that ventilation will naturally provide air to specific rooms through air movement and exchange. This study implemented 12 variations in ventilation, namely variations with cross ventilation and non-cross ventilation, then varied the size of the ventilation and the number of ventilations. This research was carried out based on simulations using Computational Fluid Dynamics. After carrying out the simulation, it was obtained that the most significant decrease in indoor temperature was when cross ventilation was applied at night with ventilation measuring 30 cm x 70 cm with a total of 4 vents, namely with a decrease in the average indoor temperature of 1.614C. After that, a cost-benefit analysis was carried out to compare the costs and benefits of a project; a cost-benefit ratio of 1.26 was obtained, where the profits were more significant than the expenditure costs, a payback period value of 0.402 was obtained, or the capital could be returned for five months, and savings were obtained. Electricity consumption costs up to year five amount to IDR 24,340,306.

Analyzing the Importance of Electronic Data Capture (EDC) for Sales Volume Prediction at Bank Merchants with the Arima Method

Payments with Electronic Data Capture (EDC) are becoming increasingly attractive to banks due to their efficiency, convenience and potential for data-based insights. The study aims to explore the use of transaction data from debit and credit cards at one of the largest banks in Indonesia to predict following year’s sales volume and growth using the Autoregressive Integrated Moving Average (ARIMA) method. Historical data on debit card and credit card transactions based on a predictive model using the ARIMA method was generated. The purpose of this study was to forecast sales volumes for the following year and identify potential growth patterns. Data preprocessing, model selection, and evaluation to ensure robust and accurate predictions as the methodology is used. The output of this study in the form of predictions for the following year can be used by banks EDC machines issuers in designing business plans in order to increase sales transactions. For further research, the results of this study can be used to measure the gap between forecasting results and actual results.

Optimal path of China's economic structure and energy demand to carbon neutrality

The changing economic structure may impact future trends in energy demand and carbon emissions since various economic production sectors have differentiated energy dependencies. Compared to consumer goods sector, investment goods sector usually has higher energy intensity, and its energy is weakly substitutable with other input factors. This paper constructs an energy-economy dynamic optimization model (EED) with heterogeneous energy dependences between investment and consumer goods sectors, and applies it to analyze the structural transformation of the Chinese economy and energy prospect. The optimization model generates a drop of investment share in GDP to 27 % in 2060, 15 percentage points lower than in 2020. Along the above transformation path, the peak demand for energy is expected to be between 6.1 and 6.3 billion (tce) in the baseline scenario, and overall energy intensity by 2060 will be 54 % lower than in 2020. The forecasting results also subject to uncertainties including technological progress. Moreover, we find that macroeconomic policy to promote consumption can help alleviate energy and environmental pressures. Thus, the development of the Chinese economy in the future needs to combine multiple channels such as promoting consumption and improving investment efficiency.

Estimating the number of foreign tourists using artificial intelligence algorithm and analyzing the socio-economic impact on the community

Abstract The high interest of foreign tourists to a region will affect the socio-economic indicators of the community. This study aims to forecast the number of foreign tourists to West Nusa Tenggara (NTB) using the backpropagation-relevance vector machine (BP-RVM) algorithm with the last 10 years of data. Forecasting using architecture is 12-25-7-3-1, activation function is radbas-logsig-logsig-purelin, training function is trainrp, learning rate is 0.1, momentum is 0.9, and RBF gamma is 0.01. The forecasting results show that in 2024 it is predicted that there will be 986,109 foreign tourists with a monthly average of 82,176, a decrease of 3.67%. The impact of foreign tourist arrivals in NTB (1) over the past three years has been significant and multidimensional, including quality of life, living standards, education, migration, public health, contribution of the tourism sector, regional income, and investment, (2) infrastructure development such as roads, bridges, and airports increased, (3) tourist demand encouraged the growth of MSMEs, (4) vocational training and education was organized, (5) the education curriculum in several schools was adjusted to the needs of the tourism industry, (6) the tourism sector contributed 10.62% to NTB’s GRDP and generated IDR 10 trillion in foreign exchange, (7) regional income from tourism reached IDR 2.5 trillion which was used for infrastructure development and public services, and (8) investment in the tourism sector created new jobs, reduced unemployment, and increased local community income. The implementation of these research findings suggests that to sustain and enhance the tourism sector’s contribution to the economy of West Nusa Tenggara, local government efforts should focus on infrastructure improvements, vocational training development, and curriculum adjustments to meet the tourism industry’s needs.

Long-term forecasting for climate classification using hybrid backpropagation relevance vector machine algorithm

Abstract Climate forecasting and classification are essential in planning cropping patterns. Therefore, the purpose of this study is to evaluate the condition of hydro-climatological data for the last 10 years and determine the trend of the data and determine the climate classification. The forecasting process utilizes a long-term forecasting method based on a hybrid backpropagation-relevance vector machine (BP-RVM) algorithm. The architecture parameters consist of three hidden layers (36-73-38-19-1), learning rate 0.1, momentum 0.9, and RBF gamma of 0.01. The forecasting results show that the BP-RVM algorithm has an average accuracy level with a category of “high accurate forecasting”. The climate classification in the Central Lombok region, Indonesia is D4 climate (slightly dry category). The increase in rainfall volume and temperature from year to year has a strong correlation with a decrease in air humidity, wind speed, and length of sunshine. The results of this study can be utilized as a basis for planning cropping patterns and handling the water needs of crops grown in the dry season. Therefore, farmers can determine the amount of water needed from dams or watersheds due to evaporation that occurs during the growing season.

FORECASTING TOTAL ASSETS OF PT. BPD KALTIM KALTARA USING THE SINGLE EXPONENTIAL SMOOTHING METHOD

PT. BPD Kaltim Kaltara is one of the regional development banks that plays a crucial role in supporting regional economic development in East Kalimantan and North Kalimantan. The company's total assets reflect significant financial stability and growth, making it an interesting topic to analyze in the context of strategic financial planning. The purpose of this study is to use the Single Exponential Smoothing (SES) approach to forecast PT. BPD Kaltim Kaltara's total assets. In the forecasting process, alpha 0,3, alpha 0,6, alpha 0,7, and alpha 0,8 are tested to determine the best value that gives the most accurate results. Based on the forecasting accuracy analysis, the SES method with alpha = 0,7 proved to be the most optimal in predicting the company's total assets, achieving MAE = 1454272,737, MSE = 4764920751283, and MAPE = 4,0433% (excellent forecasting ability). The forecasting results show an upward trend in assets, with total assets in September 2024 estimated to reach IDR 48.440.683,75. This method provides valuable guidance in thecompany's financial strategic planning, helping to anticipate future asset developments more precisely.These forecasting results also emphasize the importance of selecting the right parameters in the forecasting model to improve prediction accuracy.

Single Exponential Smoothing Method to Predict Sales Level at Agisa Kenari Printing

This research is based on the Agisa Kenari company, because in carrying out its operations, this company faces obstacles in predicting the level of paper usage that will be sold, resulting in inefficiency in managing paper inventory, which in turn can affect the smooth production and service to customers. Therefore, an effective method is needed to predict paper usage so that the company can manage inventory better and improve operational efficiency. So a more sophisticated prediction system is needed. The system was created with Python programming using the Single Exponential Smoothing (SES) method for calculation. Agisa Kenari is one of the well-known companies in Blitar city engaged in printing, making it interesting to study. The objectives of this study are (1) to describe the application of the Single Exponential Smoothing method to predict the level of use of paper sold at Agisa Kenari Printing, and (2) to describe the accuracy testing of the Single Exponential Smoothing method in predicting the level of use of paper sold at Agisa Kenari Printing using MSE (Mean Squared Error). The approach used in this research is descriptive and quantitative, while the data collection techniques used are observation, interviews, and literature studies. The results of this study are expected to provide practical and applicable solutions for companies in overcoming these obstacles and supporting better decision making. Historical sales data of various paper types, including Sidu brand A4 HVS Paper, Sidu brand F4 HVS Paper, and Sidu brand 80 gram A4 HVS Paper, are used as the basis for forecasting. The forecasting results for July show a demand of 100.02 reams for Sidu brand HVS A4 Paper with a Mean Squared Error (MSE) of 1949.9, 96.2 reams for Sidu brand HVS F4 Paper with an MSE of 1663.9, and 9.8 reams for Sidu brand HVS A4 80 gram Paper with an MSE of 11.0. These results show that the SES method can provide predictions about future paper needs, so that it can help in stock planning and more efficient operations at Agisa Kenari Printing. The implementation of forecasting using the Python program allows for a faster process, simplifying analysis and decision making in inventory management

Deep Learning and Market Trends: Building Future Economic Forecasting Models

Economic market forecasting involves predicting future market conditions based on various data. By analyzing the market trends, organizations can make informed decisions, and strategize for potential economic shifts. Accurate forecasting is crucial for minimizing risks and maximizing opportunities. In this research, we aim to develop an innovative future economic forecasting model through a deep learning (DL) algorithm. For this, we propose a novel Drosophila Food Search-drivenGate Adjusted Long Short-Term Memory[Formula: see text]DFS-GA-LSTM[Formula: see text] for analyzing the market trends and providing extensive forecasting results of future economics. We obtained a dataset comprising historical financial market data and economic indicators from various media sources. It includes stock prices and economic growth metrics to train and evaluate the proposed model. Data normalization is utilized to pre-process the gathered raw information. Kernel principle component analysis (kernel-PCA) is utilized to remove the crucial features from the processed data. In our proposed model, the DFS optimization algorithm iteratively fine-tunes the GA-LSTM architecture to enhance forecasting accuracy. The established model is implemented using Python software. During the outcome analysis phase, we evaluate our model’s performance across several parameters like RMS (0.19), RMSE (0.009), MAE (0.00011), MSE 0.003, and MAPE (26.7%). In addition, we conducted comparative analyses using diverse existing methodologies. The result reveals the superiority and efficiency of the suggested future economic forecasting model.

Peramalan Jumlah Penerimaan Mahasiswa Baru Fakultas Teknik UKI Toraja dengan Metode Single Exponential Smoothing

This study examines the effectiveness of the Single Exponential Smoothing method in forecasting the number of new student admissions at the Faculty of Engineering, Christian University of Indonesia (UKI) Toraja. This study analyzes historical data on student admissions for 11 academic years, from 2013/2014 to 2023/2024, covering four study programs, namely Mechanical Engineering, Civil Engineering, Informatics Engineering, and Electrical Engineering. The research methodology includes problem identification, data collection, analysis using the Single Exponential Smoothing method, and accuracy testing with Mean Squared Error (MSE). This study tested three values of alpha (α) = 0.1, 0.5, and 0.9. The results showed that the use of alpha = 0.9 resulted in more accurate forecasting for all study programs. This level of accuracy is validated through the lowest MSE scores, namely Mechanical Engineering of 1450.61 (prediction of 159 students), Civil Engineering of 10890.23 (prediction of 147 students), Informatics Engineering of 4332.86 (prediction of 247 students), and Electrical Engineering of 674.66 (prediction of 49 students). The comparative graph analysis of the forecast results with the actual data is consistent showing that alpha = 0.9 produces the trend closest to the actual data for all study programs. The practical implications of this study include the potential to improve the accuracy of capacity planning, more efficient resource allocation, and improve the quality of engineering at the Faculty of Engineering, UKI Toraja. These results also highlight the importance of proper selection of alpha parameters in the Single Exponential Smoothing method to optimize forecasting accuracy.

Управление результативностью микрофинансовых организаций предпринимательского финансирования с использованием алгоритмов машинного обучения

Article discusses theoretical problems of performance management of microfinance organizations as a hybrid form of entrepreneurship with dual objectives. The paper presents scoring models of decision-making on granting loans by public microfinance organizations of entrepreneurial finance based on forecasting the growth of borrowing companies. The models are built on the basis of data from 252 SME companies that received financial support from state microfinance organizations in Primorsky and Khabarovsk Krai for the period 2019-2022. The results demonstrate the high accuracy of the obtained models and their performance is confirmed by out-of-sample forecasting results. The theoretical result of the study is the development of an approach that allows integrating the principles of strategic management with the concept of value creation of public microfinance organizations. The practical significance stems from the development of a flexible decision-making algorithm that allows the management of microfinance organizations to adjust the criteria for selecting borrowers in accordance with the organizations’ strategic objectives. The algorithm assumes an ensemble structure of decision-making, where the manager of a microfinance organization receives an automated decision, but remains responsible for its result.