Us_Pathway

Data Science Internship at Futurense Technologies

🚀 Overview

This repository contains the analysis, insights, and machine learning models developed during a Data Science internship at Futurense Technologies (June–September 2024). The primary objective was to analyze business data, build predictive models, and provide actionable insights to improve business strategies.

---

🎯 Goals

• Analyze business data to uncover patterns and trends.
• Build predictive models for customer behavior and campaign performance.
• Optimize operations like resource allocation and customer retention.
• Provide insights through reports and dashboards.

---

📁 Project Components

1. Data Collection

Collected datasets included:

• Campaign Data: Spend, impressions, clicks, and platform details.
• Leads Data: Graduation details, work experience, and lead behavior.
• Phone Metrics: Call records and communication data.
• Application Tracker: Lead status throughout the process.

---

2. Data Preparation

Key preprocessing steps:

• Data Cleaning: Handled missing values and outliers.
• Feature Engineering: Derived variables like customer scores.

---

3. Exploratory Data Analysis (EDA)

• Trends in campaign effectiveness, customer behavior, and sales data.
• Platform-specific CTR analysis (Facebook, Google, LinkedIn).
• Seasonal and regional performance variations.

---

4. Machine Learning Models

Campaign Spend Prediction (CampaignML)

• Objective: Predict total campaign spend using impressions and clicks data.
• Best Model: Random Forest Regressor.
  • R²: 0.71
  • MAE: 263.11

Lead Quality Prediction (LeadML)

• Objective: Classify leads into quality levels for prioritization.
• Best Model: Random Forest Classifier.
  • Accuracy: 85%
  • F1-Score: 0.86

---

5. Pipeline Architecture

Built an automated data pipeline using Azure Services:

1. Data Extraction: Automated data retrieval from Google Sheets.
2. Data Storage: Azure Data Lake for raw and cleaned data.
3. Data Cleaning: Executed in Azure Databricks using Jupyter Notebooks.
4. Model Execution: Scalable ML models run on Azure Databricks.
5. Deployment: Real-time predictions via Azure Web App Services.

---

6. Tools and Technologies

• Programming: Python (pandas, NumPy, Scikit-learn).
• Visualization: Matplotlib, Seaborn, Power BI.
• Cloud: Azure Data Factory, Data Lake, Databricks.
• Deployment: Flask apps integrated with Azure Blob Storage.

---

🔍 Key Insights

• Platform Performance:

  • Google had the highest CTR.
  • Facebook accounted for the highest spending, while LinkedIn had the highest CPC.

• Seasonal Trends:

  • Lead generation and conversions peaked during March–April.
  • Conversion times improved consistently over the year.

• Counselor Performance:

  • Zareen achieved the fastest lead conversions.
  • Shashwat showed the highest average conversion time (>70 days).

---

🌐 Web Applications

1. CampaignML Web App

• Input: Impressions, clicks.
• Output: Predicted campaign spend.
• Tech Stack: Flask, Bootstrap, Azure Web App Services.

2. LeadML Web App

• Input: Lead features.
• Output: Predicted lead success likelihood.
• Tech Stack: Flask, Bootstrap, Azure Web App Services.

---

📊 Power BI Dashboard

Key visualizations include:

• Lead generation and payment trends.
• Campaign ROI analysis.
• Counselor performance tracking.

---

🏆 Conclusion

This project leveraged Azure services, machine learning, and visualization tools to automate workflows and provide actionable insights. The solutions developed have helped optimize business strategies and improve operational efficiency.

---

🛠️ How to Use This Repository

1. Clone the repository.
2. Install the required dependencies:

   pip install -r requirements.txt