AI Assistants for Data Analysis

Integrating AI Assistant with the REAAAPP Workflow

Kamarul Imran Musa
drki.musa@gmail.com

Universiti Sains Malaysia

Learning Objectives

  1. Understand the Data Analysis Workflow (REAAAPP)
  2. Explore AI Assistants and their capabilities
  3. Learn how to integrate AI into your workflow

Part 1

The REAAAPP Workflow

The Seven Steps

R - Reading Data

E - Exploring Data

A - making Assumptions

A - Analyzing data

A - Assessing results

P - Presenting results and interpretations

P - Publishing results, codes, dataset

Workflow Visualization

%%{init: {'theme':'base', 'themeVariables': { 'fontSize':'20px'}}}%%
flowchart LR
    A[Read] --> B[Explore] --> C[Assume] --> D[Analyze] --> E[Assess] --> F[Present] --> G[Publish]
    
    style A fill:#e74c3c
    style B fill:#f39c12
    style C fill:#f1c40f
    style D fill:#2ecc71
    style E fill:#3498db
    style F fill:#9b59b6
    style G fill:#1abc9c

Reading → Exploring

Reading:

  • Import data
  • Handle formats
  • Verify integrity

Exploring:

  • Visualize distributions
  • Check patterns
  • Identify outliers and errors
  • Flag missing or inconsistent values

Assuming → Analyzing

Assumptions:

  • Distribution (Normal, Poisson, Bernoulli etc)
  • Independence
  • Random sampling
  • Validity of assessments

Analyzing:

  • Apply methods based on assumptions
  • Fit models
  • Test hypotheses

Assessing → Presenting → Publishing

Assess:

  • Check fit
  • Validate against assumptions
  • Diagnose influential points

Present:

  • Visualize
  • Report
  • Interpret

Publish:

  • Share codes
  • Share data
  • Share documents

Part 2

AI Assistants

What Are AI Assistants?

  • Large Language Models (LLMs)
  • Trained on vast text data
  • Generate code and explanations
  • Available 24/7 or on demand if trained locally
  • Continuously improving

Four Major AI Assistants

Assistant Strength Best For
ChatGPT Broad knowledge General coding
Claude Large context Complex analysis
Gemini Google integration Research
Copilot IDE integration Active coding

R Prompt Generator

A Custom Tool for Structured Prompts

  • Aligned with REAAAPP workflow
  • Reduces ambiguity
  • Improves AI responses
  • Ensures completeness
  • Saves time

Access Here

R Prompt Generator

Part 3

Integration Approach

Integration Philosophy

REAAAPP provides structure

AI Assistants provide acceleration

R Prompt Generator provides clarity

You provide expertise and additional contexts to AI assistant (to minimize hallucinations)

The Complete Integration Process

%%{init: {'theme':'base', 'themeVariables': { 'fontSize':'14px'}}}%%
flowchart LR
    A[R Prompt<br/>Generator] --> B[Generate<br/>Prompt]
    B --> C[Copy to<br/>AI]
    C --> D[Add<br/>Resources]
    D --> E[Generate<br/>Code]
    E --> F[Copy to<br/>RStudio]
    F --> G[Run &<br/>Review]
    G --> H{OK?}
    H -->|Yes| I[End]
    H -->|No| J[Modify<br/>Prompt]
    J --> D
    
    style A fill:#3498db
    style E fill:#2ecc71
    style I fill:#27ae60
    style H fill:#f39c12

Ingredients for Success

  1. REAAAPP Workflow framework
  2. Chosen AI assistant
  3. R Prompt Generator tool
  4. Quality training resources (to add to AI Assistant)
  5. Iterative mindset

Step 1: Generate Prompt

Using R Prompt Generator:

  • Navigate to the tool
  • Select REAAAPP stage
  • Fill in task description
  • Add requirements
  • Generate specific and structured R prompt
  • R Prompt Generator: Access Here

Step 2: Copy to AI Assistant

Prepare Your Request:

  • Choose appropriate AI
  • Open new conversation
  • Paste generated prompt
  • Review for completeness

Step 3: Add Resources

Context is Critical:

  • Data samples (check your data format : csv? or excel?)
  • Variable descriptions (check the variables names)
  • Data dictionary
  • Previous examples (published codes or papers)
  • Method references (R packages vignettes or Quarto documents)
  • Special requirements (Your own notes)

Add Resources or Contexts

Step 4: Generate Solution

AI Creates Your Analysis:

  • AI processes prompt
  • Considers resources
  • Generates complete solution
  • Includes explanations
  • Suggests alternatives

Step 5: Copy to RStudio

Move to Your Environment:

  • Copy generated solution
  • Open RStudio
  • Create new R script or Quarto document
  • Save R script or Quarto document

Step 6: Run the Analysis

Execute and Monitor:

  • Review before running
  • Check package availability
  • Add data
  • Run step by step or chunck by chunk
  • Watch for errors

Step 7: Evaluate Results

Check For:

  • Correct execution
  • Expected outputs
  • Reasonable values
  • Appropriate methods

Ask Yourself:

  • Does this answer my question?
  • Are assumptions met?
  • Do results make sense?
  • What’s missing?

Decision Point

Are you satisfied?

YES

✓ Document

✓ Save

✓ Continue

NO

→ Go to Step 8

→ Refine

→ Iterate

Step 8: Refine and Iterate

When Not Satisfactory:

  1. Identify the issue
  2. Modify prompt
  3. Add more context
  4. Return to Step 3
  5. Generate new solution
  6. Test again

The Iteration Cycle

%%{init: {'theme':'base', 'themeVariables': { 'fontSize':'16px'}}}%%
flowchart LR
    A[Initial<br/>Result] --> B{Good?}
    B -->|No| C[Identify<br/>Issue]
    C --> D[Refine<br/>Prompt]
    D --> E[Add<br/>Context]
    E --> F[New<br/>Solution]
    F --> B
    B -->|Yes| G[Success]
    
    style A fill:#3498db
    style G fill:#27ae60
    style B fill:#f39c12

Best Practices

Do:

  • Use structured prompts
  • Provide context
  • Review outputs
  • Test incrementally
  • Document process

Don’t:

  • Run code blindly
  • Skip review
  • Ignore warnings
  • Over-rely on AI (verify and validate)
  • Forget domain knowledge

Benefits of Integration

Efficiency:

  • Faster development
  • Quick prototyping
  • Reduced errors

Quality:

  • Consistent methods
  • Better documentation
  • Best practices

Ethical Considerations

  • Always verify AI outputs
  • Acknowledge AI assistance
  • Maintain data privacy (should you upload or attach data?)
  • Apply domain expertise
  • Consider limitations (hallucinations, biases)

The Human Element

AI Assistants are tools

You provide:

  • Domain knowledge
  • Critical thinking
  • Quality judgment
  • Final decisions

Getting Started Checklist

Summaries

  • REAAAPP provides structure
  • AI assistants accelerate work
  • R Prompt Generator ensures clarity
  • Integration requires iteration
  • Human judgment is essential

Final Thought

REAAAPP Workflow

AI Assistance

=

Efficient, Quality Analysis

Thank You

Questions?

Start with R Prompt Generator

Practice the integration

Enjoy the learning journey

Get connected

LinkedIn: Kamarul Imran Musa

Personal Website: myanalytics