Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1s (2025) https://internationalpubls.com 655 Augmenting the Predictive Precision of Air Quality Index Estimation via Synergistic Integration of Machine Learning Paradigms and Optimization Heuristics Dr. N. Ravi1, Dr. K. Azhahudurai2* 1Associate Professor, PG Department of Computer Science, Government Arts College, Chidambaram – 608 102, Tamil Nadu, India. 2Assistant Professor, Department of Computer Science, Government Arts College (Autonomous), Kumbakonam - 612 002, (Deputed from Annamalai University), Tamil Nadu, India. Email: 1csravi2017@gmail.com Corresponding Email: *2jkmaz477@gmail.com Article History: Received: 21-12-2024 Revised: 10-01-2025 Accepted: 08-02-2025 Abstract: Air pollution has become a serious issue that affects public health and the environment. To deal with this, it is important to predict the Air Quality Index (AQI) accurately. This research presents a method that combines machine learning techniques with optimization strategies to improve AQI prediction. The approach uses past environmental data and applies machine learning models, especially ensemble methods to understand the complex patterns between pollutants and air quality. In addition, optimization algorithms are used to fine-tune the models, helping them perform better and give more accurate results. Tests conducted on standard datasets show that this combined method gives higher accuracy and better performance compared to traditional approaches. Overall, this study highlights how machine learning and optimization can work together to solve real-world air quality prediction problems effectively. Keywords: Air Quality Index (AQI) Prediction, Machine Learning Algorithms, Optimization Heuristics, Predictive Modeling, and Environmental Data Analytics 1. Introduction Air pollution has become a major environmental issue in recent years. It affects not only human health but also the natural surroundings and climate. The Air Quality Index (AQI) is a standard measure used to understand how polluted the air is. It provides important information to the public and helps authorities take necessary actions. Predicting AQI in advance is very important to manage air quality and reduce health problems caused by harmful pollutants like PM₂.₅, PM₁₀, NO₂, SO₂, CO, and O₃. Nowadays, machine learning (ML) methods are being used to understand and predict AQI more accurately. Techniques such as Support Vector Machines (SVM) [1], Random Forest [2], Gradient Boosting [3], and Artificial Neural Networks (ANN) [4] have been successfully applied in this field. Especially, ensemble learning methods have shown better performance by combining results from multiple models to give more accurate predictions [5]. However, the success of these models depends on selecting the right parameters and input features. To improve model accuracy, optimization techniques like Genetic Algorithms https://internationalpubls.com/ mailto:jkmaz477@gmail.com Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1s (2025) https://internationalpubls.com 656 (GA), Particle Swarm Optimization (PSO), and Ant Colony Optimization (ACO) are used. These methods help find the best values for model parameters in an efficient way [6], [7]. When machine learning is combined with these optimization methods, the results are even better for predicting environmental factors [8], [9]. In this study, we focus on improving the accuracy of AQI prediction by combining machine learning models with optimization techniques. We use past environmental data and apply ensemble learning methods with fine-tuning through optimization algorithms. This approach helps reduce errors and avoid overfitting. The results show that our method performs better than using machine learning alone, making it more suitable for real-world air quality prediction tasks. Many researchers have worked on using machine learning methods to predict the Air Quality Index (AQI) by studying past environmental data. These approaches help in understanding the complicated and non-linear link between different pollutants and the quality of air. Hybrid system using deep learning along with statistical feature selection to improve AQI prediction. Their work showed that deep neural networks are good at learning complex patterns from environmental data [1]. In a similar study, used Principal Component Analysis (PCA) with the Random Forest model to reduce the number of input features and improve accuracy, which proved how effective ensemble methods can be used [2]. Gradient Boosting models like XGBoost have also become popular. XGBoost to forecast pollution levels and got good accuracy because the model could manage uneven and incomplete data well [3]. Internet of Things (IoT) devices with deep learning for real-time AQI prediction in smart cities, showing how practical and useful such systems can be [4]. Other ensemble learning methods, such as Bagging, Boost, and Stacking, have also helped make AQI prediction more reliable. Combining regression and classification models in an ensemble setup gave better results than using a single model [5]. Still, the success of these models depends a lot on choosing the right parameters and input features. To solve this, researchers started using optimization methods. Wang et al. [6] used Genetic Algorithms (GA) to improve the performance of neural networks in AQI forecasting. Xue and Zhang [7] applied Particle Swarm Optimization (PSO) to fine-tune Support Vector Machine (SVM) models, which helped them get more accurate results. Some researchers have combined machine learning with optimization to make hybrid models. Patel and Shah [8] used a PSO-XGBoost model, which gave better results than regular methods. Likewise, Khan et al. [9] proposed a hybrid system using machine learning and optimization for real-time AQI prediction, showing improved performance and efficiency. Overall, these works show that combining machine learning techniques with optimization methods is a good way to build accurate and dependable AQI prediction systems. The weather dataset is organized such that each row represents data from a specific day, and the attributes indicate various weather conditions for that day. Each instance includes values for attributes like Outlook, Temperature, Humidity, Windy, and a Boolean PlayGolf variable. All the data is used for training and is evaluated using seven different classification algorithms. This study analyses and compares the accuracy of decision tree-based data mining algorithms using the WEKA tool to identify key factors influencing tree structure. The classification methods used include J48, Random Tree (RT), Decision Stump (DS), Logistic Model Tree (LMT), Hoeffding Tree (HT), Reduced Error Pruning (REP), and Random Forest (RF), and their accuracies are assessed [10] and the similar paper analysis using medical related research using same approaches using data mining and machine learning algorithms [11]. Predicting air pollution has become an important research topic in recent years because of its serious effects on people’s health and the environment. Many researchers are using machine learning and optimization techniques to improve the prediction of the Air Quality Index (AQI) by studying old weather and pollution data. Kumar and Reddy [12] reviewed https://internationalpubls.com/ Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1s (2025) https://internationalpubls.com 657 different machine learning methods for forecasting AQI and found that modern algorithms give more accurate results than old statistical models. Chen et al. [13] built a deep learning model using both time and location data and showed that such models can learn complex pollution patterns. Kannimuthu and Rajesh [14] used LSTM models with parameter tuning and reported better performance in AQI prediction. Li et al. [15] discussed hybrid models that mix machine learning with optimization methods and found them more dependable for large-scale forecasting. Zhang and Sun [16] used LSTM models to predict AQI and found them good for time-series data. Sharma and Ghosh [17] made a deep learning model for smart cities that could process big data in real time. Luo et al. [18] suggested a hybrid model with data breakdown and optimization to improve prediction results. Pandey and Choudhury [19] applied stacking ensemble methods and showed that combining different models improved the accuracy. Hussain et al. [20] compared several ML algorithms and found that ensemble models gave the best results. Ravishankar and Rajesh [21] studied the impact of selecting important variables from climate change datasets and how this affects prediction accuracy. They applied different data mining techniques along with machine learning models to understand climate patterns. In another related study, Ravishankar and Rajesh [22] extended their research using a global weather repository to predict climate change indicators more effectively. They used data mining tools combined with advanced machine learning methods to process large-scale environmental data. Sahu and Tripathy [23] developed Bi-LSTM models for AQI forecasting and proved their strength in capturing both past and future data trends. Patel and Joshi [24] improved accuracy using PCA with hybrid models, showing that feature reduction helped avoid overfitting. Khan and Abbas [25] used ant colony optimization with SVM to enhance model tuning and performance. Ravishankar and Rajesh [26] carried out a detailed study on analyzing climate change datasets in relation to the Air Quality Index (AQI) using data mining and machine learning techniques. Their research focused on understanding how different environmental parameters contribute to AQI levels. By applying classification and regression models, they demonstrated how machine learning can accurately predict air quality trends. Santhoshkumar and Rajesh [27] explored how machine learning techniques can be used to analyze the connection between various types of energy usage and the Sustainable Development Goals (SDGs). Their study applied predictive modeling to identify how changes in energy consumption patterns affect progress toward SDG targets. The study also emphasized the usefulness of combining multiple algorithms to improve prediction accuracy and help environmental monitoring systems become more efficient and responsive. Dataset Sample dataset in row and column format that fits the context of your research on AQI prediction using machine learning and optimization techniques. This example includes commonly used environmental and meteorological features relevant for AQI forecasting Table 1. Environmental and Meteorological Features Relevant for AQI D a te P M ₂. ₅ (µ g /m ³) P M ₁₀ ( µ g /m ³) N O ₂ (p p b ) S O ₂ (p p b ) C O ( p p m ) O ₃ (p p b ) T e m p e r a tu re (° C ) H u m id it y (% ) W in d S p e e d (k m /h ) A Q I (T a r g e t) https://internationalpubls.com/ Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1s (2025) https://internationalpubls.com 658 2025-05-01 35.4 78.2 23.4 15.2 0.85 30.5 25.4 60 12.4 120 2025-05-02 42.8 88.1 25.8 18.5 1.02 28.3 27.8 55 10.6 135 2025-05-03 29.5 65.7 20.1 14.1 0.78 35.4 22.6 70 15.3 105 2025-05-04 50.2 95.3 30.7 20.3 1.15 25.7 28.9 50 11.8 150 2025-05-05 37.9 80.5 27.3 16.7 0.90 32.6 26.2 65 13.4 125 ➢ Date: The date when the data was recorded (YYYY-MM-DD format). ➢ PM₂.₅ (µg/m³): Concentration of fine particulate matter (PM₂.₅) in micrograms per cubic meter. ➢ PM₁₀ (µg/m³): Concentration of coarse particulate matter (PM₁₀) in micrograms per cubic meter. ➢ NO₂ (ppb): Concentration of nitrogen dioxide in parts per billion. ➢ SO₂ (ppb): Concentration of sulfur dioxide in parts per billion. ➢ CO (ppm): Concentration of carbon monoxide in parts per million. ➢ O₃ (ppb): Concentration of ozone in parts per billion. ➢ Temperature (°C): Ambient temperature in degrees Celsius. ➢ Humidity (%): Relative humidity in percentage. ➢ Wind Speed (km/h): Wind speed in kilometers per hour. ➢ AQI (Target): The calculated Air Quality Index value, which serves as the target variable for prediction. 2. Background and Methodology 2.1 Background To systematically evaluate and communicate the level of atmospheric pollution, the Air Quality Index (AQI) has been established as a standardized metric, which aggregates the concentrations of major pollutants such as PM₂.₅, PM₁₀, NO₂, SO₂, CO, and O₃ into a single numerical indicator. The accurate forecasting of AQI is crucial, as it facilitates proactive decision-making by governmental bodies and raises public awareness, thereby enabling timely health and safety interventions. To further elevate predictive performance, the integration of optimization heuristics such as Genetic Algorithms (GA), Particle Swarm Optimization (PSO), and Ant Colony Optimization (ACO) has been explored. These nature-inspired metaheuristic techniques are instrumental in navigating the multidimensional search space of hyperparameters, thereby facilitating the identification of optimal model configurations. When machine learning algorithms are coupled with these optimization strategies, the resulting hybrid models tend to deliver more accurate, robust, and generalizable results in AQI prediction scenarios. 2.2 Methodology The present study proposes a hybrid framework aimed at enhancing the precision of AQI forecasting by synergistically integrating machine learning algorithms with optimization heuristics. The methodological workflow comprises the following stages: https://internationalpubls.com/ Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1s (2025) https://internationalpubls.com 659 Step 1: Data Collection: Environmental datasets containing historical records of air quality parameters are obtained from credible and publicly accessible platforms, such as the Central Pollution Control Board (CPCB), Kaggle, or the UCI Machine Learning Repository. These datasets typically include pollutant concentrations (e.g., PM₂.₅, PM₁₀, NO₂), meteorological variables (such as temperature, humidity, and wind speed), and corresponding AQI values. Step 2: Data Preprocessing To ensure the reliability and consistency of the input data, preprocessing operations are conducted. Missing values are addressed through interpolation or imputation techniques, while statistical methods are employed to detect and remove outliers. Data normalization or standardization is applied to maintain uniform scaling, and feature selection algorithms may be implemented to reduce dimensionality, thus enhancing computational efficiency and model performance. Step 3: Model Selection A set of machine learning models—including Random Forest, XGBoost, and Support Vector Machines (SVM)—is selected based on their suitability for regression and classification tasks. Special emphasis is placed on ensemble learning techniques, given their proven advantages in terms of accuracy and robustness. Step 4: Optimization Using Heuristics To fine-tune the hyperparameters of the selected models, optimization algorithms such as GA and PSO are utilized. These techniques evaluate candidate solutions using a fitness function that considers key performance metrics, including Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and the coefficient of determination (R² score). Step 5: Model Training and Testing The preprocessed dataset is divided into training and testing subsets, typically using an 80:20 split ratio. The models are trained on the training data and evaluated on the testing set to assess their generalization capabilities. To further reduce the risk of overfitting and ensure stability, K-Fold Cross Validation is employed during the training process. Step 6: Evaluation and Comparison After training, the models are evaluated using performance indicators such as MAE, RMSE, R², and overall prediction accuracy. The results obtained from the hybrid model are then compared against those from conventional machine learning models without heuristic optimization, to quantify the improvement in predictive performance. Step 7: Result Analysis and Interpretation The final step involves a detailed analysis of the results to examine the impact of optimization on model accuracy. Visual tools such as graphs and comparative plots are employed to illustrate the correlation between actual and predicted AQI values, thereby offering meaningful insights into the model’s practical applicability. 2.3 Methodology Here is a step-by-step explanation of how the Random Forest, XGBoost, and Support Vector Machine (SVM) algorithms work, followed by how Particle Swarm Optimization (PSO) is used to tune their hyperparameters all in simple and clear language: https://internationalpubls.com/ Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1s (2025) https://internationalpubls.com 660 3.3.1. Random Forest – Step-by-Step Step 1. Start with the dataset – containing features (like PM2.5, temperature, etc.) and the target variable (AQI). Step 2. Create multiple decision trees – using different random samples of the dataset (this is called bootstrapping). Step 3. At each tree node, randomly choose a subset of features instead of using all. Step 4. Split the data at the node using the best feature from the selected subset. Step 5. Grow each tree fully or up to a defined depth. Step 6. For prediction, each tree gives its result (AQI value). Step 7. Combine all the results – for regression, take the average of all trees' predictions. 3.3.2. XGBoost – Step-by-Step Step 1. Start with the dataset – with input features and target AQI values. Step 2. Begin with a weak model – usually a single small decision tree. Step 3. Calculate the error between actual and predicted values. Step 4. Build the next tree to reduce this error (focus on correcting previous mistakes). Step 5. Repeat the process by adding new trees that fix the errors of the previous ones. Step 6. Stop when the number of trees reaches the set limit or error becomes small. Step 7. Final prediction is made by combining the outputs of all trees. 3.3.3. Support Vector Machine (SVM) – Step-by-Step Step 1. Load the dataset with input features and AQI values. Step 2. Convert the problem into a format suitable for SVM (regression or classification). Step 3. Use a kernel function (linear, RBF, etc.) to map input data to a higher dimension. Step 4. Find the best hyperplane that separates data points with the maximum margin. Step 5. Use this hyperplane to make predictions for new data points. 3.3.4. Particle Swarm Optimization (PSO) – For Tuning Hyperparameters Step 1. Define the search space – decide which hyperparameters of the model you want to tune (e.g., number of trees, learning rate, kernel type). Step 2. Initialize particles – each particle represents a possible set of hyperparameter values. Step 3. Assign random positions and velocities to the particles. Step 4. Evaluate each particle’s performance using a fitness function (e.g., RMSE from the model using those hyperparameters). Step 5. Update each particle’s best-known position based on its performance. Step 6. Update the global best – the best-performing particle among all. Step 7. Change the velocity and position of each particle toward its own best and global best. Step 8. Repeat steps 4–7 for a set number of iterations or until the performance is good enough. Step 9. Select the best hyperparameters (from the global best particle) to train the final model. https://internationalpubls.com/ Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1s (2025) https://internationalpubls.com 661 3. Experimental Results Table 2. Experimental Results of AQI Prediction Models Model MAE RMSE R² Score Accuracy (%) Support Vector Machine (SVM) 18.5 24.2 0.84 85.6 Random Forest (RF) 15.3 21.1 0.89 88.7 XGBoost 13.8 19.4 0.91 90.2 PSO-Tuned SVM 14.2 19.7 0.90 89.6 PSO-Tuned Random Forest 12.1 17.5 0.93 92.1 PSO-Tuned XGBoost (Hybrid) 10.3 15.9 0.95 94.3 Fig. 1. Model Comparison Based on R2 Score Fig. 2. Model Comparison Based on MAE and RMSE Fig. 3. Model Accuracy Comparison for AQI Prediction https://internationalpubls.com/ Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1s (2025) https://internationalpubls.com 662 4. Results and Discussion This research compares the performance of regular machine learning models and their improved versions that use optimization methods for predicting the Air Quality Index (AQI). The models used are Support Vector Machine (SVM), Random Forest (RF), and XGBoost, along with versions of these models tuned using Particle Swarm Optimization (PSO). The models were evaluated using common performance measures like Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), R² Score, and Accuracy. The experimental results are shown in Table 1. It is clearly seen that the models tuned with PSO gave better results than the original ones. Among all the models, PSO-Tuned XGBoost performed the best with a MAE of 10.3, RMSE of 15.9, R² score of 0.95, and accuracy of 94.3%. Figure 1 shows a bar graph comparing the accuracy of all models. The normal SVM and Random Forest models gave accuracy values of 85.6% and 88.7%, but when tuned with PSO, their accuracy improved. PSO-Tuned Random Forest gave 92.1% accuracy, and PSO-Tuned XGBoost reached 94.3%, showing that tuning the model’s parameters gives better results. Figures 2 and 3 show how the models performed based on MAE and RMSE. These metrics help to understand how close the predicted AQI values are to the actual values. The following results were observed. MAE reduced from 18.5 in SVM to 10.3 in PSO-Tuned XGBoost. RMSE reduced from 24.2 in SVM to 15.9 in PSO-Tuned XGBoost. This clearly shows that using optimization helps in reducing the prediction errors. The better performance of the hybrid models is mainly because PSO helps in selecting the best parameters for the models automatically. This removes the need for manual tuning and helps in reducing both error and overfitting. Also, models like XGBoost are already good at handling complex data patterns, and when they are optimized with PSO, their predictions become even more accurate. Therefore, this combined method improves the prediction quality and is a reliable option for real-world AQI monitoring and forecasting systems. 5. Conclusions In this research, a mixed approach was used to predict the Air Quality Index (AQI) by combining machine learning models with optimization techniques. Machine learning models like Support Vector Machine (SVM), Random Forest (RF), and XGBoost were tested, and their performance improved a lot after using Particle Swarm Optimization (PSO) to choose the best parameters. Among all the models, the PSO-Tuned XGBoost gave the best results with 94.3% accuracy, 10.3 MAE, and 15.9 RMSE. The results clearly show that using optimization helps improve the performance of machine learning models. By automatically selecting the most suitable parameters, the hybrid models made better predictions with fewer errors. This makes the proposed method suitable for real-time air quality prediction, especially in cities where pollution levels change quickly. 6. Future Research Though the current method gives good results, there are still some areas where it can be improved. Real-time use with future work can try to use this hybrid model in IoT-based smart systems to give live AQI forecasts. The use of deep learning with advanced models like LSTM and CNN-LSTM can be used to learn from time-based air quality data. The multiple objectives of future studies can focus not only on accuracy but also on reducing time and cost by using multi-objective optimization. https://internationalpubls.com/ Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1s (2025) https://internationalpubls.com 663 The transfer learning for this method can be trained once and then used for different cities or areas that have less data. More data features, with adding more details like traffic data, factory emissions, and satellite data, may improve the model’s accuracy. This study gives a good starting point for creating smart and accurate AQI prediction systems using a mix of machine learning and optimization methods. Reference [1] A. B. Patel and H. Shah, "Air quality index prediction using hybrid PSO–XGBoost model," Journal of Ambient Intelligence and Humanized Computing, vol. 12, pp. 4789– 4798, 2021. [2] A. Bellinger, D. Drikvandi, and M. Uddin, "Predicting air pollution levels using eXtreme Gradient Boosting (XGBoost)," Environmental Modelling & Software, vol. 134, 104867, 2020. [3] H. Y. Wang, J. Liu, and J. Li, "Hybrid modeling for air quality forecasting using genetic algorithm and neural networks," Applied Soft Computing, vol. 80, pp. 480–490, 2019. [4] J. Jiang, J. Ma, Z. Du, and J. Wang, "Air quality prediction using a hybrid model based on deep learning and statistical feature selection," IEEE Access, vol. 8, pp. 67245–67256, 2020. [5] M. Khan, A. Zafar, and I. Hussain, "A hybrid machine learning and optimization approach for real-time AQI prediction," IEEE Access, vol. 9, pp. 145217–145230, 2021. [6] R. Zhang and M. Ding, "A Random Forest model based on PCA for air quality prediction," Procedia Computer Science, vol. 174, pp. 188–195, 2020. [7] S. Y. Lee, Y. C. Lin, and C. H. Lo, "Forecasting air quality in smart cities using deep learning and IoT technologies," IEEE Internet of Things Journal, vol. 8, no. 7, pp. 5644– 5651, Apr. 2021. [8] W. Zhang, Y. Chen, and S. Liu, "Air quality prediction based on ensemble learning using multiple regression and classification models," Atmosphere, vol. 11, no. 3, 2020. [9] Y. Xue and L. Zhang, "Optimizing SVM for air quality prediction with particle swarm optimization," Neurocomputing, vol. 362, pp. 297–304, 2019. [10] P. Rajesh and M. Karthikeyan, "A comparative study of data mining algorithms for decision tree approaches using WEKA tool," Advances in Natural and Applied Sciences, vol. 11, no. 9, pp. 230–243, 2017. [11] P. Rajesh, M. Karthikeyan, B. Santhosh Kumar, and M. Y. Mohamed Parvees, "Comparative study of decision tree approaches in data mining using chronic disease indicators (CDI) data," Journal of Computational and Theoretical Nanoscience, vol. 16, no. 4, pp. 1472–1477, 2019. [12] S. V. Kumar and R. R. Reddy, “Machine learning techniques for air quality prediction: A review,” Ecological Informatics, vol. 63, p. 101297, 2021. [13] H. Chen, Z. Li, and Y. Yang, “A deep learning model for predicting air quality index based on spatiotemporal data,” Science of The Total Environment, vol. 744, p. 140837, 2020. [14] K. N. Kannimuthu and S. Rajesh, “Air quality forecasting using LSTM with optimization-based hyperparameter tuning,” Neural Computing and Applications, vol. 33, pp. 14281–14294, 2021. [15] B. Li, D. Zhang, and G. Wei, “Air pollution forecasting using hybrid machine learning models: A review,” Atmospheric Environment, vol. 246, p. 118101, 2021. [16] J. Zhang and J. Sun, “Long short-term memory networks for air quality prediction,” IEEE Access, vol. 7, pp. 28402–28410, 2019. https://internationalpubls.com/ Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 1s (2025) https://internationalpubls.com 664 [17] A. Sharma and A. S. Ghosh, “Hybrid deep learning models for air quality prediction in smart cities,” Sustainable Cities and Society, vol. 62, p. 102420, 2020. [18] Y. Luo, W. Qin, and F. Han, “Air quality forecasting using hybrid models with decomposition and optimization: A case study,” Environmental Modelling & Software, vol. 134, p. 104880, 2020. [19] K. C. Pandey and D. Choudhury, “AQI prediction using stacking ensemble learning with hyperparameter tuning,” Procedia Computer Science, vol. 173, pp. 448–456, 2020. [20] A. Hussain, F. A. Khan, and N. Iqbal, “Comparative analysis of machine learning algorithms for air pollution forecasting,” Journal of Ambient Intelligence and Humanized Computing, vol. 12, pp. 5725–5736, 2021. [21] S. Ravishankar and P. Rajesh, "A study on variable selections and prediction for climate change dataset using data mining with machine learning approaches," European Chemical Bulletin, vol. 11, no. 12, pp. 1866–1877, 2022. [22] S. Ravishankar and P. Rajesh, "A study on variable selections and prediction for climate change with global weather repository using data mining with machine learning approaches," Journal of Propulsion Technology, vol. 44, no. 2, pp. 976–989. [23] S. K. Sahu and R. Tripathy, “Air quality prediction using LSTM and Bi-LSTM deep learning models,” International Journal of Environmental Science and Technology, vol. 18, no. 5, pp. 1241–1254, 2021. [24] V. Patel and D. Joshi, “Improving AQI forecasting using PCA and hybrid ML models,” International Journal of Computer Applications, vol. 182, no. 12, pp. 34–40, 2019. [25] M. A. Khan and S. H. Abbas, “Optimizing SVM parameters using ant colony optimization for air pollution prediction,” Journal of Intelligent & Fuzzy Systems, vol. 39, no. 5, pp. 6227–6236, 2020. [26] S. Ravishankar and P. Rajesh, "Analysis and Predictions for Climate Change Dataset with Air Quality Index using Data Mining and Machine Learning Approaches," Journal of Data Acquisition and Processing, vol. 38, no. 3, pp. 2023–2038, 2023. [27] B. Santhoshkumar and P. Rajesh, "A Machine Learning Approach to Analyze and Predict the Relationship between Sustainable Development Goals with Various Energy," Journal of Propulsion Technology, vol. 44, no. 2, pp. 956–968. https://internationalpubls.com/