A Novel Approach to Predict Flight Fares in India Using SOM-LSSVM Model Approach

Travel ticket prices can rise or fall on a regular basis depending on factors including travel schedule, destination, and duration. A prediction model will be generated in the proposed system by using machine learning techniques to the collected flight history data. Since there isn't enough information for anyone to make informed predictions regarding the future of airline ticket pricing, customers don't always know when to buy their tickets. When it comes to data preprocessing, feature selection, and training models, sequence is key. We cleaned the data and converted it from string to date time objects as part of the preprocessing; after that, we retrieved numerical values from the altered times, dates, and locations of departure and arrival. An essential part of this study was the use of EHO to optimize and select features. Feature selection plays a vital role in SOM-LSSVM model training. In every test, the proposed method beats the state-of-the-art algorithms, including SOM and LSSVM. The results clearly showed a tremendous improvement with an accuracy percentage of 97.41%.