Quickstart

1

Initialize the Client

First, import and initialize the ViewAI client with your API key:

initialize_client.py
from viewai_client import ViewAIClient

# Initialize with API key
client = ViewAIClient(api_key="your-api-key-here")

# Or use environment variable
import os
client = ViewAIClient(api_key=os.getenv("VIEWAI_API_KEY"))

Hint — Verify Connection:

verify_connection.py
# Check if the API is reachable
health = client.health.check_connection()
if health.healthy:
    print(f"✓ Connected successfully ({health.response_time_ms:.0f}ms)")
else:
    print(f"✗ Connection failed: {health.message}")
2

Prepare Your Data

The SDK works with pandas DataFrames. Example to create a sample dataset:

create_sample_dataset.py
import pandas as pd
import numpy as np

# Create sample customer churn dataset
np.random.seed(42)
n_samples = 1000

data = {
    'age': np.random.randint(18, 70, n_samples),
    'income': np.random.randint(30000, 150000, n_samples),
    'tenure_months': np.random.randint(1, 60, n_samples),
    'monthly_charges': np.random.uniform(20, 200, n_samples),
    'total_purchases': np.random.randint(0, 50, n_samples),
    'support_calls': np.random.randint(0, 10, n_samples),
    'contract_type': np.random.choice(['monthly', 'annual', 'biannual'], n_samples),
    'payment_method': np.random.choice(['credit', 'debit', 'paypal'], n_samples),
    # Target variable
    'churn': np.random.choice([0, 1], n_samples, p=[0.7, 0.3])
}

df = pd.DataFrame(data)
print(f"Dataset shape: {df.shape}")
print(f"Churn rate: {df['churn'].mean():.2%}")
3

Set Up Workspace

Workspaces organize your models. Get the default workspace or list accessible workspaces:

workspaces.py
# Get default workspace
workspace = client.retrieve_default_workspace()
print(f"Using workspace: {workspace.workspace_name}")

# Or list all accessible workspaces
workspaces = client.list_accessible_workspaces()
for ws in workspaces:
    print(f"  - {ws.workspace_name} (ID: {ws.workspace_id})")

# Set current workspace
client.set_current_workspace(workspace)
4

Train a Model

Initiate a training job on the ViewAI platform:

train_model.py
# Initiate training job
print("Starting model training...")

training_job = client.training_service.initiate_training_job(
    dataset=df,
    target_column='churn',
    workspace=workspace,
    model_name='Customer Churn Predictor',
    description='Predicts customer churn based on usage patterns',
    wait_for_completion=True  # Wait for training to complete
)

if training_job and 'completed' in training_job.status:
    print(f"✓ Training completed!")
    print(f"  Model ID: {training_job.job_id}")
    print(f"  Dashboard: {training_job.dashboard_url}")
    model_id = training_job.job_id
else:
    print("✗ Training failed")

Training Output example:

Training Progress: 100%|██████████| 30/30 [02:15<00:00]
✓ Training completed! View model at: https://app.viewai.ca/dashboard/model-123
5

Make Predictions

Single prediction example:

single_prediction.py
# Prepare test data (single customer)
customer_data = {
    'age': 35,
    'income': 75000,
    'tenure_months': 24,
    'monthly_charges': 85.50,
    'total_purchases': 15,
    'support_calls': 2,
    'contract_type': 'monthly',
    'payment_method': 'credit'
}

# Make prediction
prediction = client.prediction_service.execute_single_point_prediction(
    data=customer_data,
    model_id=model_id
)

if prediction:
    print("Prediction results:")
    print(f"  Raw prediction: {prediction.get_raw_prediction()}")
    print(f"  Probabilities: {prediction.get_probabilities()}")

Output example:

Prediction results:
  Raw prediction: {'predicted_churn': 1, 'probability_0': 0.32, 'probability_1': 0.68}
  Probabilities: {'probability_0': 0.32, 'probability_1': 0.68}

Batch prediction example:

batch_prediction.py
# Prepare batch data (multiple customers)
batch_customers = [
    {'age': 25, 'income': 45000, 'tenure_months': 6, ...},
    {'age': 50, 'income': 95000, 'tenure_months': 48, ...},
    {'age': 33, 'income': 67000, 'tenure_months': 18, ...}
]

# Or use a DataFrame
batch_df = df.drop('churn', axis=1).head(10)

# Make batch prediction
batch_job = client.prediction_service.execute_batch_prediction(
    data=batch_df,
    model_id=model_id,
    wait_for_completion=True
)

if batch_job and batch_job.status == 'completed':
    print(f"✓ Batch prediction completed!")
    print(f"  Job ID: {batch_job.job_id}")
    print(f"  View results: {batch_job.dashboard_url}")
6

Monitor Your Model

Inspect registry and run diagnostics:

monitor_model.py
# Get model from registry
models = client.registry.list_models()
our_model = next((m for m in models if m.get('dashboard_id') == model_id), None)

if our_model:
    print(f"Model: {our_model.get('dashboard_name')}")
    print(f"  Type: {our_model.get('type')}")
    print(f"  Status: {our_model.get('status')}")
    print(f"  Created: {our_model.get('created_at')}")

# Run diagnostics
diagnostics = client.health.run_diagnostics()
for check_name, result in diagnostics.items():
    status = '✓' if result.healthy else '✗'
    print(f"{status} {check_name}: {result.message}")

Complete Example

quick_start_complete.py
from viewai_client import ViewAIClient
import pandas as pd
import numpy as np
import os

# Initialize client
client = ViewAIClient(api_key=os.getenv("VIEWAI_API_KEY"))

# Verify connection
health = client.health.check_connection()
if not health.healthy:
    print("Failed to connect to ViewAI API")
    exit(1)

# Create sample data
np.random.seed(42)
df = pd.DataFrame({
    'age': np.random.randint(18, 70, 1000),
    'income': np.random.randint(30000, 150000, 1000),
    'tenure_months': np.random.randint(1, 60, 1000),
    'churn': np.random.choice([0, 1], 1000, p=[0.7, 0.3])
})

# Get workspace
workspace = client.retrieve_default_workspace()
client.set_current_workspace(workspace)

# Train model
print("Training model...")
training_job = client.training_service.initiate_training_job(
    dataset=df,
    target_column='churn',
    workspace=workspace,
    model_name='Quick Start Model',
    wait_for_completion=True
)

if not training_job or 'completed' not in training_job.status:
    print("Training failed!")
    exit(1)

model_id = training_job.job_id
print(f"Model trained successfully: {model_id}")

# Make prediction
test_data = {
    'age': 35,
    'income': 75000,
    'tenure_months': 24
}

prediction = client.prediction_service.execute_single_point_prediction(
    data=test_data,
    model_id=model_id
)

print(f"Prediction: {prediction.get_raw_prediction()}")
print("✓ Quick start completed successfully!")

Common Patterns

Train and Deploy

Using Context Manager

Configuration Object

Next Steps

Learn more and try advanced features:

  • 📘 Core Concepts - Understand workspaces, projects, and models

  • 🔐 Authentication - Secure API key management

  • 🎯 Predictions Guide - Advanced prediction patterns

Try advanced features:

  • 🚀 Model Deployment

  • 📊 Batch Processing

  • 🏥 Health Monitoring

Examples:

  • 💼 Complete Workflows

  • 🎓 Training Examples

  • 🔮 Prediction Examples

Troubleshooting

Connection Issues
Authentication Errors
Training Timeout

Best Practices

  • Always check connection before operations

  • Use environment variables for API keys

  • Handle errors gracefully with try-except blocks

  • Set appropriate timeouts for long operations

  • Monitor your models regularly

  • Use workspaces to organize projects

  • Validate data before predictions

  • Clean up resources with context managers

Support

Need help?

Ready to build production ML pipelines? Check out the Complete Workflows guide!

PreviousAuthenticationNextWorkspace and Projects

Was this helpful?