Are you curious about what makes gamers tick? 🤔 Do you want to know how to predict their engagement levels? 📊 Well, you're in the right place! 😊 This project aims to predict the engagement level of gamers based on various factors such as age, gender, location, game genre, playtime hours, and more.
Streamlit App_Machine Learning Tool To Predict Gaming Engagement Level
A machine learning-based application designed to analyze online gaming behavior and predict player engagement levels.
This application is intended to assist game developers in identifying players at risk of churn and creating personalized experiences to increase player satisfaction and loyalty.
- Choose the characteristics that describe the player and the game 📝.
- Select the features that are relevant to your analysis.
- Get the predicted engagement level of the player 📊.
- Use the predicted engagement level to create personalized experiences for the player.
- Use the app to predict player engagement levels for note-taking, research, or content creation 📝.
Containing 40,034 samples and 13 features. Here's a sneak peek at what's inside:
PlayerID: Unique identifier for each playerAge: Age of the playerGender: Gender of the player (Male/Female)Location: Location of the playerGameGenre: Genre of the game played (Strategy, Sports, Action, RPG, Simulation)PlayTimeHours: Number of hours playedInGamePurchases: Number of in-game purchases madeGameDifficulty: Difficulty level of the game (Easy, Medium, Hard)SessionsPerWeek: Number of sessions played per weekAvgSessionDurationMinutes: Average duration of each session in minutesPlayerLevel: Level of the playerAchievementsUnlocked: Number of achievements unlockedEngagementLevel: Engagement level of the player (Low, Medium, High)
Here's a step-by-step guide to how to approached this project:
- Data Preprocessing 🧹: Removing missing values, encoding categorical variables, and scaling numerical variables.
- Exploratory Data Analysis (EDA) 🔍: Exploring the dataset to understand the distribution of each feature and their relationships.
- Modeling 🤖: Training and evaluating several machine learning models to predict the engagement level of gamers. The models used include:
- Logistic Regression
- K-Nearest Neighbors
- Support Vector Machines
- Decision Tree
- Random Forest
- AdaBoost
- Gradient Boosting
- Naive Bayes
- Neural Network
- XGB
- HistGradientBoostingClassifier
- LGBMClassifier
- CatBoostClassifier
- ExtraTreesClassifier
- Hyperparameter Tuning 🔧: Tuning hyperparameters using Bayesian optimization to improve the performance of the models.
The best performing model was the XGB model with an accuracy of 91.64% on the test set.
| Model | Training Accuracy | Test Accuracy |
|---|---|---|
| XGB | 93.39% | 91.64% |
| RandomForest | 92.12% | 91.42% |
Predicting gamer engagement is a complex task, but with the right approach and techniques, we can achieve high accuracy. 📈 This project demonstrates the power of machine learning in understanding gamer behavior and predicting their engagement levels.
- Analyze online gaming behavior to identify patterns and trends.
- Predict player engagement levels to inform game development and marketing strategies.
- Use the application to identify areas for improvement in game design and player experience.
This app is licensed under AGPL-3.0 License and is for personal use only and should not be used for commercial purposes. The GamerDNA model is a pre-trained model and may not always produce accurate results.
I hope you found this project informative and engaging! 😊
If you're interested in collaborating and contributing to the project, please let me know! I'd love to hear from you.
To get started with this project, you'll need to:
- Install the required libraries, including TensorFlow, Keras, and OpenCV 📦
- Download kaggle datasets using
download -d gpiosenka/butterfly-images40-species📈 - Run the code to train and evaluate the model 🤖
Enjoy working with the content! 😊