intelligems-segment-analysis（智能细分市场分析）

/intelligems-segment-analysis

Analyze segment-level performance of active Intelligems experiments.

Shows which segments (device, visitor type, traffic source) each experiment is winning in, with:

• Visitors and orders per segment
• Which variation is performing best
• RPV lift and confidence levels
• GPV lift (if COGS data exists)

Prerequisites

• Python 3.8+
• Intelligems API key (get from Intelligems support)

Step 1: Get API Key

Ask the user for their Intelligems API key.

If they don't provide it upfront, ask:

"What's your Intelligems API key? You can get one by contacting support@intelligems.io"

IMPORTANT: Never use a hardcoded or default API key. The user must provide their own.

Step 2: Set Up Project

Create a project directory with the analysis script:

mkdir -p intelligems-segment-analysis
cd intelligems-segment-analysis

Create config.py

Copy from references/config.py:

# Intelligems Segment Analysis Configuration

# API Configuration
API_BASE = "https://api.intelligems.io/v25-10-beta"

# Thresholds (Intelligems Philosophy: 80% is enough)
MIN_CONFIDENCE = 0.80  # 80% probability to beat baseline
MIN_RUNTIME_DAYS = 14  # Don't make status judgments until test runs 2+ weeks

# Segment types to analyze
SEGMENT_TYPES = [
    ("device_type", "BY DEVICE"),
    ("visitor_type", "BY VISITOR TYPE"),
    ("source_channel", "BY TRAFFIC SOURCE"),
]

Create segment_analysis.py

Copy the full script from references/segment_analysis.py.

Create.env

Create .env file with the user's API key:

INTELLIGEMS_API_KEY=<user's key here>

Install Dependencies

pip install requests python-dotenv tabulate

Step 3: Run Analysis

python segment_analysis.py

Step 4: Interpret Results

Status meanings:

• Doing well - Variant beating control with 80%+ confidence
• Not doing well - Control beating variant with 80%+ confidence
• Inconclusive - Not enough confidence either way
• Too early - Test running less than 2 weeks (don't trust yet)
• Low data - Not enough orders to calculate confidence

Example output:

======================================================================
  Homepage Price Test
   Runtime: 21 days | Visitors: 45,000 | Orders: 320
======================================================================

📱 BY DEVICE
╭─────────┬──────────┬────────┬───────────┬─────────────┬──────────┬────────────╮
│ Segment │ Visitors │ Orders │ Variation │ Status      │ RPV Lift │ Confidence │
├─────────┼──────────┼────────┼───────────┼─────────────┼──────────┼────────────┤
│ Mobile  │   32,000 │    180 │ +5%       │ Doing well  │ +12.3%   │ 87%        │
│ Desktop │   13,000 │    140 │ +5%       │ Inconclusive│ +4.2%    │ 62%        │
╰─────────┴──────────┴────────┴───────────┴─────────────┴──────────┴────────────╯

This shows the +5% price variant is winning on mobile (87% confidence) but inconclusive on desktop.

Troubleshooting

"INTELLIGEMS_API_KEY not found"

• Ensure .env file exists with the key
• Or export: export INTELLIGEMS_API_KEY=your_key

"No active experiments found"

• Check that experiments have status "started" in Intelligems dashboard

"Error fetching experiments"

• Verify API key is correct
• Check network connection

References

• references/segment_analysis.py - Full analysis script
• references/config.py - Configuration file
• references/setup-guide.md - Detailed setup instructions