1. Project Overview
2. Data Sources
   • Walmart Sales Data
   • Amazon Sales Data FY2020-21
   • Amazon Customer Reviews
3. Features
4. Modeling Approach
   • Sales Forecasting Models
     • Prophet Model
     • SARIMA Model
     • Random Forest Regressor
   • Sentiment Analysis Model
5. Installation & Setup
6. Usage Guide
   • Data Upload
   • Making Predictions
   • Interpreting Results
7. Real-World Applications
8. Educational Applications
9. App Breakdown

The Retail Analytics Suite is a comprehensive data analysis and visualization platform that combines sales data from multiple retail sources to provide actionable insights and predictive analytics. The application analyzes Walmart sales data, Amazon sales data, and Amazon customer sentiment data to deliver a holistic view of retail performance and customer satisfaction.

• Source: Walmart store sales records
• Period: 2020-2021
• Format: CSV
• Update Frequency: Weekly
• Key Metrics: Weekly sales, store information, and economic indicators
• Link: Walmart Dataset on Kaggle

• Source: Amazon sales records
• Period: 2020-2021 fiscal year
• Format: CSV
• Update Frequency: Daily
• Key Metrics: Order details, product information, and pricing data
• Link: Amazon Sales Dataset on Kaggle

• Source: Amazon customer feedback
• Period: 2020-2021
• Format: CSV
• Update Frequency: Daily
• Key Metrics: Review content, ratings, and customer sentiment
• Link: Amazon Reviews Dataset on Kaggle

1. Sales Analysis

   • Time series visualization
   • Seasonal trend analysis
   • Store performance comparison
   • Revenue forecasting

2. Customer Sentiment Analysis

   • Review sentiment scoring
   • Word cloud visualization
   • Category-wise sentiment analysis
   • Trend analysis of customer satisfaction

3. Price Analytics

   • Price distribution analysis
   • Discount impact assessment
   • Category-wise pricing strategies
   • Competitive pricing analysis

4. Interactive Dashboards

   • Dynamic filtering
   • Real-time calculations
   • Custom date range selection
   • Multiple visualization options

• Purpose: Long-term sales forecasting with seasonal decomposition
• Implementation:
  • Multiplicative seasonality model
  • Yearly, weekly, and daily seasonality components
  • Holiday effects incorporation
  • Automatic changepoint detection

Advantages:

• Handles missing data effectively
• Captures multiple seasonality patterns
• Robust to outliers
• Automatic trend changepoint detection

• Purpose: Short to medium-term sales forecasting
• Implementation:
  • Order selection: (1,1,1) for trend
  • Seasonal order: (1,1,1,7) for weekly patterns
  • Rolling validation approach

Advantages:

• Captures both trend and seasonality
• Provides confidence intervals
• Strong statistical foundation
• Good for short-term predictions

• Purpose: Feature-based sales prediction
• Implementation:
  • 200 trees with max depth of 15
  • Minimum samples split: 5
  • Minimum samples leaf: 2

Advantages:

• Handles non-linear relationships
• Feature importance ranking
• Robust to outliers
• Good for complex pattern recognition

• Purpose: Customer review sentiment scoring
• Implementation:
  • VADER sentiment scoring
  • Custom text preprocessing
  • Weighted sentence analysis

Advantages:

• Specifically tuned for social media text
• Handles emoji and punctuation
• Multiple sentiment dimensions

# Clone the repository
git clone 'https://github.com/Diparna/Sales_Forecasting/'

# Install required packages
pip install -r requirements.txt

# Run the Streamlit app
streamlit run app.py

• Format Requirements: Specify CSV format
• Supported File Types: CSV files only

• Step-by-step guide for selecting data and running models
• Input parameters explanation

• Understanding the visualizations
• Reading the forecasts

• Predict demand for products to optimize inventory and avoid stockouts or overstocking
• Plan promotional campaigns based on forecasted sales trends

• Anticipate demand spikes to adjust production schedules and logistics planning
• Improve supplier relations by sharing accurate demand forecasts

• Help businesses allocate budgets effectively by forecasting revenue streams
• Aid in financial planning for startups or small businesses

• Predict seasonal trends in bookings, helping with staff planning and resource allocation
• Forecast customer preferences for targeted marketing

• Assist restaurants and cafes in purchasing raw materials by forecasting sales volumes
• Predict trends in customer dining habits for menu optimization

• Plan production schedules based on sales forecasting to reduce waste
• Align resources with expected demand for better cost management

• Demonstrate time-series analysis, ARIMA models, and machine learning applications in forecasting
• Visualize the impact of data preprocessing and model selection on predictive accuracy

• Allow students to upload their datasets to practice forecasting in real-world scenarios
• Provide hands-on experience with sales datasets for project-based learning

• Illustrate the power of predictive analytics in a simple, user-friendly manner
• Engage visitors by allowing them to tweak parameters and see how forecasts change

• Use the app as a sandbox environment for students to explore retail analytics
• Simulate scenarios to teach resource allocation based on demand predictions

• Serve as a tool for teaching Streamlit, Python, and data visualization techniques
• Show how to integrate data science with app development to build impactful tools

• Default landing page with an overview of the three main analyses

1. Store Timeline
2. Sales Forecasting
3. Original Dataset Overview
4. Engineered Features Overview
5. Feature Distribution
6. Correlation Analysis
7. Seasonal Sales Analysis
8. ARIMA Forecasting
9. Summary

1. Sales Prediction
2. IDA (Initial Data Assessment)
3. Data Transformation
4. Product Overview
5. Pricing Analysis
6. Rating Analysis

1. Product Recommendations
2. Sentiment Distribution Tab:
   • Distribution analysis
   • Summary statistics
3. Detailed Analysis Tab:
   • Sentiment score distribution
4. Rating Analysis Tab:
   • Sentiment vs rating visualization
   • Correlation metrics
5. Word Clouds Tab:
   • Word clouds for positive, neutral, and negative reviews