---

DataMining Project

This project is a collection of data mining tasks implemented in Java, leveraging machine learning and pattern mining techniques. It includes a series of applications, each focused on a specific data mining task, including association rule mining, classification, and frequent pattern analysis.

Project Structure

DataMining/
├── .vscode/                 # VS Code configurations (launch and settings)
├── bin/                     # Compiled Java class files
├── datasets/                # Datasets used for data mining tasks
│   ├── bank.arff
│   ├── bank_no.arff
│   ├── bank_yes.arff
│   ├── COVID19.arff
│   └── News.arff
├── lib/                     # External libraries
│   ├── spmf-1.jar           # SPMF library for frequent pattern mining
│   └── weka.jar             # Weka library for machine learning tasks
├── output/                  # Output files (e.g., pattern mining results)
└── src/                     # Java source files
    └── com/davidlangat/datamining/
        ├── BankAssociationRulesAnalysis.java
        ├── BankPatternMiner.java
        ├── BankPatternMiningComparison.java
        ├── ClassificationTask.java
        ├── ClassificationTask2.java
        ├── MaximumPatternMiner.java
        └── NewsClassification.java

Getting Started

Prerequisites

• Java JDK (version 1.8 or above)
• Visual Studio Code with the Java Extension Pack (recommended for running the project in VS Code)

Libraries

This project uses the following external libraries:

• SPMF for frequent pattern mining
• Weka for machine learning algorithms

Ensure that spmf-1.jar and weka.jar are placed in the lib directory as per the settings.json configuration.

Setup

1. Clone the repository and open the DataMining folder as the workspace in Visual Studio Code.
2. Open the terminal and navigate to the project’s root directory.
3. Run the following command to compile the project:

   javac -d bin -cp "lib/*" src/com/davidlangat/datamining/*.java

4. You can then run each main class individually using the VS Code launch.json configurations.

Usage

Each Java file in the src/com/davidlangat/datamining directory corresponds to a specific data mining task. You can run these files individually using the configurations in .vscode/launch.json.

Data Mining Tasks

1. BankAssociationRulesAnalysis.java

Description: This class performs association rule analysis on a banking dataset using the SPMF library. It aims to extract valuable insights regarding banking customer behaviors.

• Input: datasets/bank.arff
• Output: Association rules based on defined confidence and support thresholds.

2. BankPatternMiner.java

Description: BankPatternMiner finds frequent patterns within the banking dataset, allowing analysts to understand common transactions and trends.

• Input: datasets/bank.arff
• Library: Uses SPMF library for pattern mining.

3. BankPatternMiningComparison.java

Description: This class compares the performance of Apriori and FP-Growth algorithms on different support thresholds, enabling an evaluation of efficiency between the two methods.

• Input: datasets/bank.arff
• Output: Outputs a comparison of runtime and pattern count for each algorithm.

4. ClassificationTask.java

Description: ClassificationTask performs multi-class classification on a dataset, using algorithms like J48 (decision trees), PART, Naive Bayes, and OneR.

• Input: datasets/COVID19.arff
• Library: Weka
• Output: Accuracy and evaluation metrics for each classifier.

5. ClassificationTask2.java

Description: A variant of ClassificationTask that includes cost-sensitive classification. It incorporates a cost matrix to handle imbalanced data and improve model performance on critical classes.

• Input: datasets/COVID19.arff
• Library: Weka
• Output: Cost-sensitive evaluation metrics.

6. MaximumPatternMiner.java

Description: This class mines maximum patterns in the banking dataset, which are the largest frequent itemsets, providing a summary of common transaction sets.

• Input: datasets/bank_yes.arff and datasets/bank_no.arff
• Output: Top patterns for both "yes" and "no" classes.

7. NewsClassification.java

Description: NewsClassification is a text classification task that uses various classifiers like IBk (k-nearest neighbors), SMO (SVM), J48, and Hoeffding Tree to categorize news articles.

• Input: datasets/News.arff
• Library: Weka
• Output: Classification accuracy and evaluation results for each classifier.

Configuration Files

launch.json

The launch.json file in .vscode is set up to allow easy execution of each main class. Each configuration points to one of the data mining tasks in the project.

{
  "version": "0.2.0",
  "configurations": [
    // Configurations for each main class
  ]
}

settings.json

The settings.json file defines the source path (src), output path (bin), and external libraries (lib/weka.jar and lib/spmf-1.jar), which are required for compilation and execution.

{
  "java.project.sourcePaths": ["src"],
  "java.project.outputPath": "bin",
  "java.project.referencedLibraries": [
    "lib/weka.jar",
    "lib/spmf-1.jar"
  ]
}

Running the Project

1. In VS Code:

   • Open the Run and Debug tab.
   • Select a configuration (e.g., "Run BankPatternMiner") and press Start Debugging.

2. From the Command Line:

   • Use java -cp "bin;lib/*" com.davidlangat.datamining.ClassName to run any specific class, replacing ClassName with the desired Java file.

Troubleshooting

• FileNotFoundException: Ensure that dataset paths in each Java file match the relative paths (e.g., datasets/bank.arff).
• Library Issues: Verify that both spmf-1.jar and weka.jar are in the lib folder.

Contributing

Contributions are welcome! If you have ideas to extend or improve the project, please fork the repository and submit a pull request.

License

This project is licensed under the MIT License. See the LICENSE file for more details.

---