📡 Neural Network SMS Text Classifier

This project trains a binary classification neural network to distinguish between spam and ham (non-spam) SMS messages using TensorFlow and Keras.

---

📁 Dataset

The data used is the SMS Spam Collection dataset from FreeCodeCamp, consisting of two .tsv files:

• train-data.tsv — training set
• valid-data.tsv — validation/test set

Each line is formatted as:

<label> <TAB> <message>

Where <label> is either "ham" or "spam".

---

⚙️ Installation & Setup

bash pip install tensorflow pip install tensorflow-datasets pip install pandas numpy matplotlib

!wget https://cdn.freecodecamp.org/project-data/sms/train-data.tsv
!wget https://cdn.freecodecamp.org/project-data/sms/valid-data.tsv

---

🧠 Model Architecture

• Input: Raw SMS text string
• TextVectorization Layer: tokenizes and pads the input
• Embedding Layer: converts tokens to dense vectors (dim=16)
• GlobalAveragePooling1D: reduces sequence
• Dense Layer (24 units, ReLU): learns non-linear combinations
• Dropout (0.5): for regularization
• Output Layer (sigmoid): predicts probability of spam

---

🏋️ Training Parameters

• Loss: BinaryCrossentropy
• Optimizer: Adam
• Epochs: 20
• Metrics: Accuracy
• Validation Set: valid-data.tsv

---

🔮 Prediction Function

This function accepts a raw string message and returns:

[probability (0–1), "ham" or "spam"]

Example:

predict_message("how are you doing today?")
# Output: [0.0123, "ham"]

Code:

def predict_message(pred_text):
    input_data = np.array([pred_text], dtype=object)
    prediction = model.predict(input_data)
    spam_prob = float(prediction[0][0])
    label = 'spam' if spam_prob > 0.5 else 'ham'
    return [spam_prob, label]

---

✅ Test Pass Criteria

Your model must correctly classify the following test messages:

[
  "how are you doing today",
  "sale today! to stop texts call 98912460324",
  "i dont want to go. can we try it a different day? available sat",
  "our new mobile video service is live. just install on your phone to start watching.",
  "you have won £1000 cash! call to claim your prize.",
  "i'll bring it tomorrow. don't forget the milk.",
  "wow, is your arm alright. that happened to me one time too"
]

All predictions must match expected labels ("ham" or "spam") for success.

---

Colab Notebook Link (version problems)

https://colab.research.google.com/drive/1s7oLDkVU_kBd1biHSb7nmPi6Coq1gkze?authuser=2#scrollTo=lMHwYXHXCar3

🏆 Certificate

Upon passing the test suite, FreeCodeCamp awards a certificate for this project as part of their Machine Learning curriculum.

---