brunaaaz/_previsaode_reclamacoescom_modelos_supervisionados_

📊 Customer Retention Analytics Dashboard

Project Status: Active Type: Predictive Analytics / Business Intelligence

💼 Business Context

In the retail sector, customer retention is critical. This project provides a data-driven solution to predict customer complaints and churn risk. By leveraging Machine Learning algorithms on historical data, this dashboard allows analysts to identify risk factors and take proactive measures.

🎯 Key Features

• —Data Exploration: Interactive filtering by demographics (Education, Marital Status) to analyze specific customer segments.
• —Predictive Modeling: Real-time training of classification models (Random Forest, Gradient Boosting) to predict the Complain target.
• —Risk Driver Analysis: Automated identification of the most significant variables influencing customer dissatisfaction (e.g., Recency, Income).
• —Performance Metrics: Detailed breakdown of model accuracy, recall (sensitivity), and AUC-ROC curves.

🛠️ Tech Stack

• —Python: Core language.
• —Streamlit: Web application framework.
• —Scikit-Learn: Machine learning pipeline and metrics.
• —Imbalanced-Learn (SMOTE): Handling class imbalance in the dataset.
• —Plotly: Interactive visualizations.

🚀 How to Run

1. 1.Clone this repository.
2. 2.Install dependencies:

    pip install -r requirements.txt

1. 1.Launch the application:

    streamlit run app.py

📈 Insights & Results

The analysis highlights that customer engagement metrics (such as Recency of purchase) and demographic factors (Marital Status) are strong predictors of dissatisfaction, allowing the Customer Success team to prioritize interventions.