Analysis Modes Guide

Quick Start

The AI Pattern Analyzer now supports four analysis modes that let you balance speed and accuracy:

# Fast mode - Quick analysis of document start (default)
python analyze_ai_patterns.py document.md

# Adaptive mode - Intelligent sampling based on document size
python analyze_ai_patterns.py document.md --mode adaptive

# Sampling mode - Custom sampling strategy
python analyze_ai_patterns.py document.md --mode sampling --samples 10 --sample-size 3000

# Full mode - Complete document analysis (most accurate)
python analyze_ai_patterns.py document.md --mode full

Mode Comparison

Mode	Speed	Accuracy	Best For	Coverage
Fast	⚡⚡⚡ Fastest	⭐⭐ Basic	Quick checks, early drafts	~2000 chars/dimension
Adaptive	⚡⚡ Fast	⭐⭐⭐ Good	Most documents, standard workflow	Adjusts to size
Sampling	⚡ Medium	⭐⭐⭐⭐ High	Large documents, custom needs	User-defined
Full	🐌 Slowest	⭐⭐⭐⭐⭐ Best	Final review, critical content	100%

Detailed Mode Descriptions

Fast Mode (Default)

When to Use:

Quick sanity checks during writing
Early draft reviews
Rapid iteration on small changes
CI/CD pipeline checks

How It Works:

Analyzes first ~2000 characters per dimension
Skips heavy computations
Provides directional feedback

Example:

python analyze_ai_patterns.py draft.md
# or explicitly:
python analyze_ai_patterns.py draft.md --mode fast

Typical Speed: 2-5 seconds for most documents

Adaptive Mode

When to Use:

Standard analysis workflow
Medium to large documents (5k-100k words)
When you want balanced speed and accuracy
Most production use cases

How It Works:

Documents < 5k chars: Full analysis
Documents 5k-50k chars: 5 samples × 2000 chars
Documents > 50k chars: 10 samples × 2000 chars
Intelligently selects representative sections

Example:

python analyze_ai_patterns.py article.md --mode adaptive

Typical Speed: 5-15 seconds depending on document size

Sampling Mode

When to Use:

Very large documents (100k+ words)
Custom sampling requirements
Performance tuning
Specific section analysis

How It Works:

You control the number of samples and sample size
Choose sampling strategy (uniform, weighted, start, end)
Optimizes for your specific needs

Example:

# Sample 8 sections of 5000 chars each
python analyze_ai_patterns.py large-doc.md --mode sampling --samples 8 --sample-size 5000

# Weighted sampling (favors start/end)
python analyze_ai_patterns.py book.md --mode sampling --samples 15 --strategy weighted

Sampling Strategies:

uniform: Even distribution across document
weighted: More samples from start/end (default)
start: Focus on beginning
end: Focus on conclusion

Typical Speed: 10-30 seconds depending on parameters

Full Mode

When to Use:

Final manuscript review
Critical content (academic, legal)
Maximum accuracy requirements
Benchmark/baseline analysis

How It Works:

Analyzes entire document
All dimensions at full depth
Most comprehensive results
No sampling or approximation

Example:

python analyze_ai_patterns.py final-manuscript.md --mode full

Typical Speed: 30-120+ seconds for large documents

Decision Matrix

Use this matrix to choose the right mode:

Situation	Recommended Mode	Why
Writing first draft	Fast	Quick feedback loop
Reviewing edits	Adaptive	Balances speed and accuracy
Pre-submission check	Full	Maximum confidence
Large book chapter	Sampling	Efficient for large content
CI/CD pipeline	Fast or Adaptive	Speed matters
Academic paper	Full	Accuracy critical
Blog post	Fast or Adaptive	Sufficient for web content
Legal document	Full	Risk mitigation

Integration Examples

Batch Analysis with Modes

# Analyze multiple files with adaptive mode
python analyze_ai_patterns.py chapter-*.md --mode adaptive --batch

# Custom sampling for large batch
python analyze_ai_patterns.py *.md --mode sampling --samples 5 --batch --output results.tsv

Detailed Analysis with Modes

# Get detailed findings with full analysis
python analyze_ai_patterns.py manuscript.md --mode full --detailed

# Quick detailed check
python analyze_ai_patterns.py section.md --mode fast --detailed --output-format json

Dual Score Reports with Modes

# Optimization report with adaptive mode
python analyze_ai_patterns.py article.md --mode adaptive --scores

# Full accuracy dual score
python analyze_ai_patterns.py critical-doc.md --mode full --scores

History Tracking with Modes

History automatically tracks which mode was used:

# First iteration with full analysis
python analyze_ai_patterns.py draft.md --mode full --save-to-history

# Quick check with fast mode
python analyze_ai_patterns.py draft.md --mode fast --save-to-history

# View history with mode information
python analyze_ai_patterns.py draft.md --history-full

The history report shows mode and analysis time for each iteration:

ITERATION 1: Initial baseline
Timestamp:     2025-01-15
Mode:          FULL
Analysis Time: 45.3s
Quality:       72.0 / 100  (GOOD)
...

Dry-Run Mode

Preview what the analyzer will do without running analysis:

# See what fast mode will analyze
python analyze_ai_patterns.py large-doc.md --mode fast --dry-run

# Preview sampling strategy
python analyze_ai_patterns.py book.md --mode sampling --samples 10 --dry-run

Output shows:

Mode selected
Coverage statistics
Dimensions that will be analyzed
Estimated analysis time

Coverage Statistics

Display detailed coverage information:

python analyze_ai_patterns.py document.md --mode adaptive --show-coverage

Output shows:

Document size
Characters analyzed
Coverage percentage
Mode-specific details

Performance Tuning Tips

For Speed

Use Fast Mode for iterations:

python analyze_ai_patterns.py draft.md --mode fast

Reduce sample count:

python analyze_ai_patterns.py large.md --mode sampling --samples 3

Use adaptive for automatic optimization:

python analyze_ai_patterns.py *.md --mode adaptive --batch

For Accuracy

Use Full Mode for final checks:

python analyze_ai_patterns.py final.md --mode full

Increase sampling for large documents:

python analyze_ai_patterns.py book.md --mode sampling --samples 20 --sample-size 5000

Use weighted sampling strategy:

python analyze_ai_patterns.py manuscript.md --mode sampling --strategy weighted

For Large Documents

Start with Sampling Mode:

python analyze_ai_patterns.py large-book.md --mode sampling --samples 15

Adjust sample size based on document:
- Documents < 50k words: 5-10 samples
- Documents 50k-100k words: 10-15 samples
- Documents > 100k words: 15-25 samples

Use adaptive if document size varies:

python analyze_ai_patterns.py *.md --mode adaptive --batch

Mode Help

Get detailed information about modes:

python analyze_ai_patterns.py --help-modes

This displays:

All four modes with descriptions
Sampling strategies
Usage examples
Performance characteristics

Troubleshooting

"Analysis taking too long"

Solution: Switch to a faster mode

# Instead of:
python analyze_ai_patterns.py huge.md --mode full

# Try:
python analyze_ai_patterns.py huge.md --mode sampling --samples 10

"Results seem inaccurate"

Solution: Use more thorough analysis

# Instead of:
python analyze_ai_patterns.py important.md --mode fast

# Try:
python analyze_ai_patterns.py important.md --mode full

"Out of memory errors"

Solution: Reduce sample size or use fast mode

# Instead of:
python analyze_ai_patterns.py massive.md --mode full

# Try:
python analyze_ai_patterns.py massive.md --mode sampling --samples 10 --sample-size 2000

"Invalid sample count"

Error: Sample count must be between 1 and 50

Solution: Adjust --samples parameter

# Valid:
python analyze_ai_patterns.py doc.md --mode sampling --samples 10

# Invalid (too high):
python analyze_ai_patterns.py doc.md --mode sampling --samples 100

"Mode not working as expected"

Solution: Use --dry-run to verify configuration

python analyze_ai_patterns.py doc.md --mode sampling --samples 5 --dry-run

Best Practices

Development Workflow

During Writing: Use Fast mode for immediate feedback
```
python analyze_ai_patterns.py draft.md --mode fast
```
After Major Changes: Use Adaptive mode for balanced review
```
python analyze_ai_patterns.py draft.md --mode adaptive
```

Before Submission: Use Full mode for final validation

python analyze_ai_patterns.py final.md --mode full --scores

Team Workflow

CI/CD Pipeline: Fast or Adaptive for speed

python analyze_ai_patterns.py *.md --mode adaptive --batch --output results.tsv

Code Review: Adaptive mode with history

python analyze_ai_patterns.py changed-files.md --mode adaptive --save-to-history

Release Validation: Full mode for critical content

python analyze_ai_patterns.py release-notes.md --mode full --detailed

Content Type Guidelines

Content Type	Recommended Mode	Frequency
Blog posts	Fast or Adaptive	Per save
Technical docs	Adaptive	Per major change
Marketing copy	Adaptive or Sampling	Daily
Academic papers	Full	Before submission
Books/ebooks	Sampling (10-15 samples)	Per chapter
Legal documents	Full	Before finalization
Social media	Fast	Per post
Email campaigns	Adaptive	Before send

Advanced Usage

Custom Sampling Strategy

# Analyze start and end heavily
python analyze_ai_patterns.py novel.md --mode sampling --samples 20 --strategy weighted

# Uniform sampling across entire document
python analyze_ai_patterns.py article.md --mode sampling --samples 8 --strategy uniform

# Focus on beginning (intro analysis)
python analyze_ai_patterns.py report.md --mode sampling --samples 5 --strategy start

# Focus on end (conclusion analysis)
python analyze_ai_patterns.py thesis.md --mode sampling --samples 5 --strategy end

Combining with Other Flags

# Full analysis with all reports
python analyze_ai_patterns.py manuscript.md --mode full --detailed --scores --save-to-history

# Fast batch with JSON output
python analyze_ai_patterns.py *.md --mode fast --batch --output-format json --output results.json

# Sampling with coverage display
python analyze_ai_patterns.py large.md --mode sampling --samples 10 --show-coverage

# Adaptive with dry-run first
python analyze_ai_patterns.py doc.md --mode adaptive --dry-run
python analyze_ai_patterns.py doc.md --mode adaptive

Mode Selection Algorithm

The tool uses this logic when mode is not specified:

Check --mode flag: If specified, use that mode
Default to Fast: Most common use case
Adaptive recommended: For production workflows
Full for critical: When accuracy is paramount

Migration from Previous Versions

If you've been using the analyzer without modes:

Old command:

python analyze_ai_patterns.py document.md

New equivalent (same behavior):

python analyze_ai_patterns.py document.md --mode fast

Recommended upgrade:

python analyze_ai_patterns.py document.md --mode adaptive

All existing commands work without changes. Modes are purely additive.

FAQ

Q: Which mode should I use most often? A: Adaptive mode for most work, Fast for quick checks, Full for final reviews.

Q: Do different modes give different scores? A: Scores may vary slightly due to sampling, but trends and insights are consistent.

Q: Can I change modes between iterations in history? A: Yes! History tracks which mode was used for each iteration.

Q: Does batch mode work with all analysis modes? A: Yes, all modes work with --batch flag.

Q: Will modes affect my existing scripts? A: No, existing commands work identically. Default is Fast mode (previous behavior).

Q: How do I know which mode was used? A: Check the output header, history report, or use --show-coverage.

Q: Can I use modes with --detailed flag? A: Yes, modes work with all existing flags (--detailed, --scores, --batch, etc.).

Q: What if I run out of memory with Full mode? A: Use Sampling mode with fewer samples or smaller sample size.

Summary

Fast: Quick checks (default)
Adaptive: Smart balance (recommended)
Sampling: Large documents
Full: Maximum accuracy

Start with Adaptive, adjust based on your needs. Use --dry-run to preview, --show-coverage to understand, and --help-modes for detailed reference.

FilesExpand file tree

analysis-modes-guide.md

Latest commit

History

analysis-modes-guide.md

File metadata and controls

Analysis Modes Guide

Quick Start

Mode Comparison

Detailed Mode Descriptions

Fast Mode (Default)

Adaptive Mode

Sampling Mode

Full Mode

Decision Matrix

Integration Examples

Batch Analysis with Modes

Detailed Analysis with Modes

Dual Score Reports with Modes

History Tracking with Modes

Dry-Run Mode

Coverage Statistics

Performance Tuning Tips

For Speed

For Accuracy

For Large Documents

Mode Help

Troubleshooting

"Analysis taking too long"

"Results seem inaccurate"

"Out of memory errors"

"Invalid sample count"

"Mode not working as expected"

Best Practices

Development Workflow

Team Workflow

Content Type Guidelines

Advanced Usage

Custom Sampling Strategy

Combining with Other Flags

Mode Selection Algorithm

Migration from Previous Versions

FAQ

Summary