Biography

is a senior academician in Mathematics with over 15 years of experience. She holds a Ph.D. in Applied Mathematics. She has published several research papers. She has guided student projects. Her research interests include numerical methods and optimization techniques. She actively participates in academic activities. She contributes to curriculum development. She is a member of professional bodies. She has received awards for teaching excellence. She mentors students in research and problem-solving. She is committed to academic and research excellence.Probabilistic models and statistical analysis are the go-to tools for forecasting flight delays. Nevertheless found that these models had difficulty handling high-dimensional data and extracting non-linear relationships. In order to anticipate flight delays it suggested a CNN-LSTM and DL system [6]. There are primarily three parts to the suggested CNN-LSTM model: For the purpose of predicting flight delays a CNN an LSTM network and finally a RFC are utilized. To begin the spatial adjustments between regions are extracted using a CNN framework [7]. The LSTM network is then used to describe temporal dynamics based on the output of the CNN algorithm. Numerous ML modelsare investigated in this study such as LR, NB, NN, RF, XGBoost, CatBoost and LightGBM [8]. Adding meteorological data to further tests using the SMOTE helps fix the data imbalance problem. This comprehensive method guarantees accurate forecasting in different environments [9]. The majority of the studies that were reviewed used RF as their method of choice. Nearly as popular are GB, ANN, and DT. The ML which DL is a subset allows computers to construct complicated ideas using simpler representations [10]. The DL is able to tackle intricate problems by describing each layer of the model as a nested simple mapping which simplifies the complicated mapping [11]. For complicated data patterns DL provides a wealth of advanced frameworks such as LSTM and CNN [12]. Images and data with a defined grid-like layout are ideal for the analysis and processing tasks that CNN are designed to handle. In contrast LSTM and NN excel in handling data sequences [13]. It appears that DL is a practical and effective method for extracting features from data and predicting future flight delays caused by weather given that weather has both spatial and temporal components. Utilizes LSTM and BiLSTM models to provide a categorization method for flights delays [14]. Flight delays are a big problem for airlines since they are inconvenient for customers and cost money. A classification test has demonstrated encouraging results for the powerful DL approaches of BiLSTM and LSTM models. For this research it trained and tested the BiLSTM and LSTM models using a dataset of flight on-time performance data obtained from the of Transportation Statistics and utilized in the research [15]. However, in order to analyze the performance comparative similarities with others it is highly significant to use a BiLSTM type of RNN model with structure and an offline dataset such that used for flight delays [16]. Finally improve the SVM model for flight delay prediction it adds the previous knowledge of flight delay as an inequality matrix [17]. Using the flight records for the month and distance employed the CatBoost method to estimate flight delays based on weather data for variables including wind speed and visibility. Flight delay prediction using several ML models such as KNN, SVM, NBC and RF. Through extensive investigation into feature selection, pre-processing, and model design, they attempted to tackle the multiple obstacles associated with flight delay prediction networks. Here are some important points that our research brings to light: I propose extracting features from complicated flight delay data using the XGBoost model and then assigning relevance scores to those features. Set different feature priority levels to obtain different amounts of features, and then use numerous metrics to evaluate the best feature set. The MIX_LSTM model was designed with the goal of achieving higher performance in flight delay prediction and detection. It is based on the classic DL LSTMmodel. III. PROPOSED SYSTEM Accurately predicting when planes will take off allows for better use of airport support, apron, and runway resources, which in turn facilitates better team decision-making.To meet the increasing demand for transportation in civil aviation, more flights will need to be operated. Flight operations are required to make use of the apron, the airport's assistance, and the runway resources. Flight operations may experience delays due to the disparity between increasing flight demand and available resources. The Kaggle dataset was the first source for the dataset. The data came from the United States Department of Transportation's Bureau of Transportation Statistics [18]. A total of 1,685,597 flights from January 2019, February 2019, and January 2020 make up the training dataset. A dataset consisting of 569,133 flights in February 2020 is used for testing purposes.