Nixtla TimeGPT-2 Migrator

Facilitates a smooth transition from TimeGPT-1 to TimeGPT-2.

Purpose

Automates the migration process from TimeGPT-1 to TimeGPT-2, identifying compatibility issues and generating updated code.

Overview

Evaluates existing TimeGPT-1 workflows for compatibility with TimeGPT-2. Identifies potential breaking changes and suggests necessary code modifications. Employs API checks and data schema validation. Provides updated code snippets and configuration examples for TimeGPT-2. Generates a migration report summarizing the changes required.

Prerequisites

Tools: Read, Write, Edit, Glob, Grep

Environment: NIXTLA_TIMEGPT_API_KEY

Packages:

pip install nixtla pandas matplotlib statsforecast pyyaml

Instructions

Step 1: Analyze codebase

Scan the codebase for TimeGPT-1 API usage patterns using the analysis script.

Script: {baseDir}/scripts/analyze_codebase.py

Usage:

python {baseDir}/scripts/analyze_codebase.py /path/to/your/codebase

The script searches for:

• timegpt.forecast() calls
• timegpt.create_model() calls
• timegpt.load_data() calls
• timegpt.train() calls

Output: analysis_report.txt listing all TimeGPT-1 usage instances.

Step 2: Run compatibility check

Execute the compatibility checker to validate data schema and identify unsupported features.

Script: {baseDir}/scripts/compatibility_check.py

Usage:

python {baseDir}/scripts/compatibility_check.py --data sample_data.csv

The script validates:

• Data schema (unique_id, ds, y columns)
• Data types (datetime for ds, numeric for y)
• API availability
• Unsupported TimeGPT-1 features

Output: migration_report.txt with compatibility assessment.

Step 3: Apply migration changes

Use the migration script to update your codebase with TimeGPT-2 compatible code.

Script: {baseDir}/scripts/apply_migration.py

Usage:

python {baseDir}/scripts/apply_migration.py main.py

The script performs automatic replacements:

• timegpt.forecast() → client.forecast()
• from timegpt import TimeGPT → from nixtla import NixtlaClient
• timegpt = TimeGPT() → client = NixtlaClient(api_key=...)
• Removes deprecated timegpt.create_model() calls
• Updates data schema conversion code

Important: Review all changes before committing to version control.

Step 4: Generate TimeGPT-2 configuration

Create a TimeGPT-2 configuration file with recommended settings.

Script: {baseDir}/scripts/generate_config.py

Usage:

python {baseDir}/scripts/generate_config.py

Output: timegpt2_config.yaml with configuration parameters.

Output

• analysis_report.txt: Summary of TimeGPT-1 usage in codebase.
• migration_report.txt: Compatibility assessment and migration plan.
• updated_codebase/: Modified source code with TimeGPT-2 compatible calls (after applying changes).
• timegpt2_config.yaml: Configuration file for TimeGPT-2.

Error Handling

1. Error: TimeGPT-1 API endpoint not found Solution: Ensure TimeGPT-1 API is accessible or skip API validation step.

2. Error: Incompatible data schema Solution: Update data input format to match TimeGPT-2 requirements (unique_id, ds, y columns).

3. Error: Missing API Key Solution: Set the NIXTLA_TIMEGPT_API_KEY environment variable.

4. Error: Unsupported TimeGPT-1 feature Solution: Refactor code to use equivalent TimeGPT-2 functionality or alternative approaches.

5. Error: File not found during migration Solution: Verify the file path and ensure the file exists before running the migration script.

Examples

Example 1: Basic Migration

Before (TimeGPT-1):

from timegpt import TimeGPT
timegpt = TimeGPT()
forecast = timegpt.forecast(data, h=24)

After (TimeGPT-2):

from nixtla import NixtlaClient
import os
client = NixtlaClient(api_key=os.getenv('NIXTLA_TIMEGPT_API_KEY'))
forecast = client.forecast(df=data, h=24, freq='H')

Example 2: Configuration Update

Before (config.json):

{
  "model": "timegpt-1",
  "horizon": 24
}

After (timegpt2_config.yaml):

api_key: YOUR_API_KEY_HERE
model_name: TimeGPT-2
frequency: H
forecast_horizon: 24
data_format: Nixtla

Example 3: Full Migration Workflow

# Step 1: Analyze codebase
python {baseDir}/scripts/analyze_codebase.py ./my_project

# Step 2: Check compatibility
python {baseDir}/scripts/compatibility_check.py --data ./data/sample.csv

# Step 3: Apply migration (review migration_report.txt first)
python {baseDir}/scripts/apply_migration.py ./my_project/main.py

# Step 4: Generate config
python {baseDir}/scripts/generate_config.py

Resources

• TimeGPT-2 API documentation: https://docs.nixtla.io/
• Migration guide: https://docs.nixtla.io/docs/migration-guide
• NixtlaClient reference: https://nixtlaverse.nixtla.io/nixtla/
• Scripts: {baseDir}/scripts/ directory contains all migration tools