The AkinatorFB application is designed to interactively predict football players based on a series of user-provided responses to questions about player attributes. The program employs a decision-making algorithm to narrow down the dataset iteratively, We are using the latest 2024 version of FC24 dataset.
The AkinatorFC application is structured in a clear and modular manner, adhering to the best practices in code layout and design, The code is organized into a class-based structure for code organisation facilitates code readability, maintainability ,and reusability.
LOAD THE DATASET: The model starts by loading a dataset all_players.csv was downloaded from (Stefano Leone, 2023), which is initially being trimmed to essential columns needed for the guessing, Then the trimmed dataset is saves as trimmed_dataset.csv, Then it is printed. INITIALISE VARIABLES: Track asked questions, Then the path taken during the game and a set of for players are being declared to save.
The game is limited to ask 20 questions, asking for the users response to confirm or deny specific attributes of the football player.
Based on the user response (yes/no/probably(P)/probably-no(PN)t/I don’t know (idk)), the dataset is filtered to narrow down the potential player matching the given criteria.’
A ‘Probably (P)’ option is included to capture user uncertainty and a ‘Probably-not (PN) option additional logic filtering player aged 25 and above to proceed cautiously.
If only one player remains in the dataset, the model makes a prediction. If the prediction is accurate, the game ends, and user feedback is requested. If not, the model learns from the inaccurate prediction.
In order to predict accurately and not to repeat the questions for making more user interactive, Question are being asked randomly from the available columns and entries which are taken from a sample of the dataset.
The program comes with an additional feature called Best Guess, it is when the user completes 20 question and the model wasn’t able to narrow down to one player. The model makes the best guess based on the remaining dataset.
PROS: INTERACTING AND ENGAGING : The game format makes it interactive and engaging for users. Clear instructions and prompts contribute to a positive user experience. RANDOMIZED QUESTION SELECTION: The use of randomization in selecting questions adds an element of unpredictability, making each session potentially unique. This contributes to user engagement and replay value. ERROR HANDLING: The application includes error handling mechanisms to manage invalid user inputs, ensuring the program can gracefully recover from unexpected scenarios. CONS: LIMITED DATASET EXPLORATION: The application only allows users to explore 20 questions in the dataset, which may limit their ability to accurately predict answers to more complicated queries. A more adaptive approach that adjusts the number of questions based on the dataset size could be beneficial. 20 QUESTIONS MAX: The game has a 20 question maximum, which might not be the best option in every situation.
DYNAMIC QUESTION LIMIT: To ensure flexibility, introduce a dynamic question limit that is dependent on the size of the dataset. MACHINE LEARNING INTEGRATION: Incorporate machine learning algorithms for dynamic learning from user feedback. USER FEEDBACK ANALYSIS: Implement more characterised analysis of user feedback, considering partial correctness and incorporating probabilistic predictions.
In conclusion, The implementation of features that improve adaptability, learning capacity and user satisfaction, the proposed enhancements seek to elevate AkinatorFC24 model. While this improvements comes with challenges, they open the door to creating a more sophisticated and intelligent player prediction game.
Stefano Leone. (2023, October). EA Sports FC 24 complete player dataset. Retrieved from Kaggle: https://www.kaggle.com/datasets/stefanoleone992/ea-sports-fc-24- complete-player-dataset