Skip to main content

Workflow

Overview​

BI Analyst provides a conversational approach to creating data visualizations. Users upload their data files and interact with the system through natural language to generate and refine charts.

Step-by-Step Workflow​

Step 1: Data Upload​

  • File Upload: Upload your CSV or Excel file to the platform
  • Supported Formats: CSV and Excel (.xlsx, .xls) files only
  • Data Processing: System automatically parses your file and analyzes the data structure
  • Data Validation: Basic validation ensures data quality for visualization

Step 2: Chart Request​

Advanced Model Processing

User describes visualization requirements and receives AI-generated chart suggestions

  • Natural Language: Describe the chart you want using everyday language
  • Be Specific: Include details about what data columns to use and chart type preferences
  • Context Provided: Mention the purpose or audience for the visualization

Example Requests:

"Create a bar chart showing sales by region"
"Show monthly revenue trends as a line chart"
"Generate a scatter plot of price vs quantity sold"
"Make a pie chart of product categories by revenue"

Step 3: Chart Generation​

  • SAM Processing: The SAM framework analyzes your request and data
  • Chart Selection: System determines the most appropriate visualization type
  • Professional Styling: Automatic application of clean formatting and colors
  • Chart Creation: Generate the visualization based on your specifications

Step 4: Chart Interaction​

  • Hover Details: Move your mouse over chart elements to see specific data values
  • Data Exploration: Click on different parts of the chart to explore data points
  • Visual Analysis: Use the chart to identify patterns and insights in your data

Step 5: Chart Modification​

Advanced Model Processing

User modifies chart styling, data filters, and chart types through natural language requests

All modifications are done through additional prompts to the system:

Style Changes:

  • "Change the bars to blue color"
  • "Make the chart title larger"
  • "Add data labels to the bars"

Data Adjustments:

  • "Filter to show only Q4 data"
  • "Group the data by month instead of week"
  • "Show only the top 10 categories"

Chart Type Changes:

  • "Convert this to a horizontal bar chart"
  • "Show this data as a pie chart instead"
  • "Create a line chart with the same data"

Step 6: Export and Use​

  • PNG Export: Download your finished chart as a high-resolution PNG image
  • Professional Quality: Charts are suitable for presentations, reports, and documents
  • Ready to Use: Downloaded images can be used immediately in your business materials

Getting Started Tips​

Effective Chart Requests​

  • Be Clear: Specify which columns to use for x-axis, y-axis, and grouping
  • State Purpose: Mention if it's for a presentation, report, or analysis
  • Include Filters: Specify any data filtering requirements upfront

Working with Data​

  • Clean Data: Ensure your data is properly formatted with clear column headers
  • Data Types: Make sure numeric columns contain numbers and date columns are formatted correctly
  • Column Names: Use descriptive column headers for better chart labeling

AI Prompt Recommendations​

When you start working with new data, the system provides:

  • Suggested Prompts: Examples of charts that work well with your data structure
  • Data Insights: Basic information about what the system found in your data
  • Chart Recommendations: Suggestions for effective ways to visualize your specific dataset

Chart Modification Through Prompts​

Filtering Data​

  • Use prompts to focus on specific time periods, categories, or value ranges
  • Request comparisons between different segments of your data
  • Ask for top/bottom performers or specific data subsets

Drilling Down​

  • Request more detailed views of specific data points
  • Ask for breakdowns by additional dimensions
  • Explore different time periods or categories through follow-up questions

Style and Format Changes​

  • Modify colors, fonts, and layout through conversational requests
  • Adjust chart titles, axis labels, and legends
  • Request different chart orientations or arrangements