Skip to content

QUICK_START.md

Latest commit

 

History

History
230 lines (178 loc) · 7.75 KB

QUICK_START.md

File metadata and controls

230 lines (178 loc) · 7.75 KB

🚀 Quick Start Guide - Employee Review Analytics Dashboard

📋 What You Have Now

I've successfully created a comprehensive employee review analytics dashboard with all the features you requested:

✅ Completed Features

🔍 Data Processing & Classification

  • Sentiment Analysis: Automatically classifies reviews as positive, negative, or neutral using VADER + TextBlob
  • User Type Classification: Categories employees as:
    • Intern (stagiaire, trainee, apprentice)
    • Senior/Manager (senior, lead, manager, director)
    • Consultant (consultant, advisor, specialist)
    • Developer/Engineer (developer, engineer, programmer, analyst)
    • Junior/New Employee (based on employment length ≤ 1 year)
    • Experienced Employee (≥ 3 years)
    • General Employee (default)

📊 Interactive Dashboard Features

  • Real-time Filters: User type, sentiment, year range, review source
  • Multiple Tabs: Overview, Topic Analysis, Word Clouds, Review Search, AI Recommendations
  • Interactive Charts: Built with Plotly for responsive visualizations
  • Keyword Tracking: Real-time alerts for specific terms

🤖 AI-Powered Recommendations

  • Gemini Integration: Uses your provided API key for intelligent analysis
  • Critical Issues Identification: Automatically finds common problems
  • Actionable Insights: Specific, implementable improvement suggestions
  • Export Capabilities: Download filtered data and summary reports

🌐 Access Your Dashboard

The dashboard is now running at: http://localhost:8501

🖥️ Browser Access

  1. Open your web browser
  2. Go to: http://localhost:8501
  3. The dashboard should load automatically

📂 Files Created

Sentiment Analysis/
├── dashboard.py              # Main dashboard application ✅
├── setup_dashboard.py       # Python setup script ✅
├── run_dashboard.bat        # Windows launcher ✅
├── requirements.txt         # Python dependencies ✅
├── README.md               # Comprehensive documentation ✅
├── merge_reviews.py        # Review merging script ✅
├── merged_reviews.json     # Your processed data ✅
└── analysis.py             # (Empty - for your additional analysis)

🎛️ How to Use the Dashboard

Left Sidebar - Filters

  • User Type: Filter by employee category
  • Sentiment: Show positive/negative/neutral only
  • Year Range: Select time periods (slider)
  • Review Source: Glassdoor vs Indeed
  • Keyword Tracking: Monitor specific terms

Main Tabs

1. 📊 Overview Tab

  • Sentiment distribution pie chart
  • Sentiment by user type bar chart
  • Sentiment trends over time
  • Rating distribution analysis
  • Key metrics at the top

2. 🎯 Topic Analysis Tab

  • Automatic topic modeling using LDA
  • Separate topics for positive/negative/neutral reviews
  • Key themes identification

3. ☁️ Word Clouds Tab

  • Visual word frequency analysis
  • Separate clouds for positive and negative reviews
  • Interactive and responsive

4. 🔍 Review Search Tab

  • Smart search functionality
  • Sort by: Recent, Rating, Helpful
  • Expandable review cards with full details
  • Pagination for large datasets

5. 💡 AI Recommendations Tab

  • Critical issues analysis from negative reviews
  • AI-generated improvement suggestions using Gemini
  • Export options for data and reports
  • Success metrics and implementation timelines

🚀 Getting Started

First Time Setup

  1. Double-click run_dashboard.bat (easiest option)

    • OR run: python setup_dashboard.py
    • OR manually: streamlit run dashboard.py

  2. Open browser to: http://localhost:8501

  3. Explore the data:

    • Start with the Overview tab to see overall sentiment
    • Use filters to narrow down specific groups
    • Check Word Clouds for visual insights
    • Generate AI recommendations for improvement actions

Key Insights You Can Get

📈 Sentiment Analysis

  • Overall satisfaction trends
  • Problem areas by user type
  • Temporal changes in sentiment
  • Source comparison (Glassdoor vs Indeed)

👥 User Type Analysis

  • Which employee groups are most/least satisfied
  • Common issues by seniority level
  • Onboarding vs experienced employee feedback

🎯 Topic Modeling

  • Automatic theme detection in reviews
  • Most discussed topics by sentiment
  • Emerging issues and positive developments

🤖 AI Recommendations

  • Specific action items for improvement
  • Priority ranking of issues
  • Success metrics for tracking progress

💡 Pro Tips

For HR Teams

  1. Use negative sentiment filter + topic analysis to identify systemic issues
  2. Track keyword alerts for terms like "management", "salary", "toxic"
  3. Export summary reports for leadership presentations
  4. Monitor trends over time to measure improvement

For Management

  1. Filter by user type to understand different employee segments
  2. Use AI recommendations for specific action planning
  3. Compare Glassdoor vs Indeed sentiment for platform differences
  4. Track rating distribution changes over time

For Data Analysis

  1. Export filtered datasets for deeper analysis
  2. Use search functionality to find specific issues
  3. Combine filters for targeted insights
  4. Monitor keyword tracking for emerging trends

🔧 Troubleshooting

Dashboard Not Loading

  • Check if the terminal shows: "You can now view your Streamlit app in your browser"
  • Try refreshing the browser page
  • Ensure port 8501 isn't blocked by firewall

Missing Data

  • Verify merged_reviews.json exists in the same directory
  • Re-run the merge script if needed

AI Recommendations Not Working

  • Check internet connection (Gemini API requires online access)
  • Verify the API key is valid
  • Try generating recommendations with fewer reviews

Performance Issues

  • Use filters to reduce dataset size
  • Close other browser tabs
  • Restart the dashboard if it becomes slow

🌟 Advanced Features

Export Capabilities

  • CSV Export: Filtered review data
  • Summary Reports: Key insights and statistics
  • Custom Analysis: Use exported data in Excel/other tools

Keyword Monitoring

  • Track specific terms in real-time
  • Get sentiment breakdown for keywords
  • Monitor emerging issues or positive developments

API Integration

  • Gemini AI for intelligent recommendations
  • Extensible for other AI services
  • Custom prompts for specific analysis needs

🎯 Next Steps

Immediate Actions

  1. Explore the dashboard with different filter combinations
  2. Generate AI recommendations for critical issues
  3. Export a summary report for stakeholders
  4. Set up keyword tracking for important terms

Regular Monitoring

  1. Weekly reviews of sentiment trends
  2. Monthly AI recommendations for new issues
  3. Quarterly reports for leadership
  4. Continuous keyword tracking for early warnings

Data Expansion

  1. Add more review sources as they become available
  2. Update the merged dataset regularly
  3. Customize user type classifications for your organization
  4. Add custom metrics specific to your needs

🎉 Congratulations!

You now have a fully functional, AI-powered employee review analytics dashboard that provides:

✅ Automated sentiment analysis
✅ Smart user categorization
✅ Interactive visualizations
✅ Topic modeling and insights
✅ AI-powered recommendations
✅ Export and reporting capabilities
✅ Real-time filtering and search
✅ Keyword tracking and alerts

Start exploring your data insights now at: http://localhost:8501 🚀

Need help? Check the comprehensive README.md file or review the code in dashboard.py for customization options.