SMS Spam Detection

This repository teaches to code an SMS Spam Classifier using Multi-Layer Perceptron. Here, scikit-learn pipelines are used for pipelining the training and testing task on the training and testing sets respectively. The codes for the same are avaialble in main.py.

A high level view of the system is as follows:

• Import libraries
• Import SMS Spam Collection dataset
• Define function for removing punctuations and stopwords
• Split training and testing set
• Create pipeline
  • Construct Bag of Words
  • Compute tf-idf scores
  • Instantiate MLP Classifier
• Fit the pipeline on the training set
• Test the pipeline on the test set
• Print the results

Action Usage

To use this action in your repository, create a file like .github/workflows/spam_detect.yml with the following content:

name: Spam Detection Action
on:
  push:
    branches: [ master ]
  pull_request:
    branches: [ master ]
jobs:
  build:
    runs-on: ubuntu-latest
    steps:
    - uses: actions/checkout@v2
    - name: SMS Spam Detection Action
      uses: amaaniqbal/sms-spam-detection@v1.21
      with:
        # Message to be checked for spam
        message: "input message to be tested goes here..."

Script Usage

Run the file main.py without any command line arguments to view the classifier performance on the available test set. In order to test on the custom input, run the following command:

python main.py "Custom input string goes here"

In the command, custom input string can also be specified without quotes.

Contributing

Please feel free to create an issue if you find a bug or send a pull request if you wish to improve the existing code/add a new feature.