Settlemate/autopilot-email-classifier

Autopilot Email Classifier

A scikit-learn TF-IDF + Logistic Regression model that predicts whether an email is purchase-related (purchase) or not (non-purchase).

Quick Usage

# 1️⃣ Clone this repository
git clone https://huggingface.co/Settlemate/autopilot-email-classifier
cd autopilot-email-classifier

# 2️⃣ Install dependencies
pip install -r requirements.txt

# 3️⃣ Load the model and classify

import joblib

# Load Model
bundle = joblib.load("classifier.pkl")
pipeline  = bundle["pipeline"]
threshold = bundle["threshold"]

# Function to classify email
def classify_email(subject: str, body: str) -> str:
    text   = f"{subject} {body}"
    proba  = pipeline.predict_proba([text])[0, 1]
    return "purchase" if proba >= threshold else "non-purchase"

# Example usage
if __name__ == "__main__":
    subj = "Solo founder, $80M exit, 6 months: The Base44 bootstrapped startup success story | Maor Shlomo"
    body = """Base44's founder on bootstrapping to profitability, using AI to write 90% of his code, why he turned down VC money, and signing an acquisition deal as missiles were flying"""
    print(classify_email(subj, body))