Cloud Integration

Xplainable Cloud

Xplainable Cloud provides enterprise-grade model deployment, collaboration, and production management capabilities through the separate xplainable-client package.

Overview

The xplainable ecosystem includes two packages:

1. xplainable - Core transparent ML package (open source)
2. xplainable-client - Cloud integration package (separate install)

This separation allows you to use the core xplainable features without cloud dependencies, while providing full cloud capabilities when needed.

Installation

1pip install xplainable-client

Cloud Package

The cloud client is completely separate from the core xplainable package. Install both for full functionality.

Quick Start

Basic Setup

1import os

2from xplainable_client import Client

3

4# Initialize the client

5client = Client(api_key=os.environ['XP_API_KEY'])

With Custom Configuration

1# See xplainable-client documentation for additional configuration options

2client = Client(

3 api_key=os.environ['XP_API_KEY']

4)

Core Features

Dataset Management

Public Datasets

Access curated datasets for learning and experimentation.

1import xplainable as xp

2

3# List available public datasets (requires xplainable-client)

4datasets = xp.list_datasets()

5print(f"Available datasets: {len(datasets)}")

6

7# Load a specific dataset

8data = xp.load_dataset('titanic')

9print(f"Dataset shape: {data.shape}")

Model Management

Model Creation

Train a model locally, then push it to the cloud.

Model Loading

Load models from the cloud for inference and inspection.

1# Train a local model

2from xplainable.core.models import XClassifier

3model = XClassifier()

4model.fit(X_train, y_train)

5

6# Create model in cloud

7result = client.models.create_model(

8 model=model,

9 model_name="Customer Churn Model",

10 model_description="Predicts customer churn",

11 x=X_train,

12 y=y_train

13)

1# Load models from cloud

2# See xplainable-client documentation for full API reference

Preprocessing Management

1# Preprocessing pipelines are available via the xplainable-preprocessing package

2# pip install xplainable-preprocessing

3

4# Save a preprocessing pipeline to cloud

5preprocessor_id = client.preprocessing.create_preprocessor(

6 preprocessor=pipeline,

7 preprocessor_name="Standard Pipeline",

8 preprocessor_description="Data preprocessing pipeline"

9)

Model Deployment

Production Deployment

Deploy models as REST APIs with one command.

1# Deploy model to production

2deployment = client.deployments.deploy(

3 model_version_id="your-version-id"

4)

Model Inference

1# Make predictions and get explanations using the local model

2predictions = model.predict(X_test)

3model.explain()

4

5# For cloud-based inference, see xplainable-client documentation

Advanced Features

The xplainable-client package provides additional cloud features including model versioning, team collaboration, and model management. Refer to the xplainable-client documentation for the full API reference.

AI Assistant Integration

AI-Powered Insights

The cloud client includes AI assistant capabilities for automated insights and report generation.

1# Generate automated report via GPT integration

2report = client.gpt.generate_report(model_id="your-model-id")

Complete Workflow Example

Here's a complete example showing the full workflow from training to deployment:

1import xplainable as xp

2from xplainable.core.models import XClassifier

3from xplainable.core.optimisation.bayesian import XParamOptimiser

4from xplainable_client import Client

5import pandas as pd

6from sklearn.model_selection import train_test_split

7import os

8

9# Initialize client

10client = Client(api_key=os.environ['XP_API_KEY'])

11

12# Load data

13data = xp.load_dataset('titanic')

14X, y = data.drop('Survived', axis=1), data['Survived']

15X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2)

16

17# Optimize hyperparameters

18optimiser = XParamOptimiser(metric='roc-auc', n_trials=30)

19params = optimiser.optimise(X_train, y_train)

20

21# Train model with optimized parameters

22model = XClassifier(**params)

23model.fit(X_train, y_train)

24

25# Explain the model

26model.explain()

27

28# Create model in cloud

29result = client.models.create_model(

30 model=model,

31 model_name="Titanic Survival Predictor",

32 model_description="Transparent model for predicting survival",

33 x=X_train,

34 y=y_train

35)

36

37# Deploy to production

38deployment = client.deployments.deploy(

39 model_version_id=result["version_id"]

40)

41

42print(f"Model deployed successfully!")

Security & Authentication

API Key Management

1# Set API key via environment variable (recommended)

2# export XP_API_KEY="your-api-key-here"

3

4# Or set programmatically (not recommended for production)

5client = Client(api_key="your-api-key-here")

Organization & Team Management

1# Initialize with organization context

2client = Client(

3 api_key=os.environ['XP_API_KEY']

4)

5

6# See xplainable-client documentation for team and organization management

Error Handling

1from xplainable_client import Client

2

3try:

4 client = Client(api_key="your-api-key")

5except Exception as e:

6 print(f"Client error: {e}")

7 # Check your API key and network connection

Best Practices

Security

• Never hardcode API keys - Use environment variables
• Use organization/team contexts for proper access control
• Regularly rotate API keys for security
• Monitor API usage through the dashboard

Performance

• Cache model objects to avoid repeated downloads
• Use batch predictions for multiple samples
• Monitor deployment metrics for performance insights
• Version models systematically for reproducibility

Collaboration

• Use descriptive model names and descriptions
• Tag models with relevant metadata
• Share models appropriately with team permissions
• Document model assumptions and limitations

Migration Guide

From Internal Client (Pre-v1.2.9)

If you were using the internal client, migrate to the separate xplainable-client package:

1# NEW (External client)

2from xplainable_client import Client

3client = Client(api_key="your-key")

The xplainable-client package is now the recommended way to interact with Xplainable Cloud. See the xplainable-client documentation for the full API reference.

Support

Need help with cloud integration?

• Documentation: Comprehensive API reference
• Community: Join our user community
• Support: Enterprise support available
• Issues: Report bugs and feature requests

Next Steps

Ready to deploy your first model? Check out our tutorials for complete examples, or explore the Python API documentation for detailed technical information.