Quick start: your own data

Time to complete: ~30–45 minutes (first time)
You need: Signed-in account, CSV or Excel (.csv / .xlsx), clear column headers

This is Path A in User journeys.

Before you start

• Sidebar access to Lake (/data)
• ML: 100+ rows, one column to predict (target)
• Forecast: 30+ rows, parseable date/time column + numeric values
• File under 100 MB (split larger exports)

Bronze — add your table (~5 min)

1. Sidebar → Data Warehouse → Lake.
2. Click Add data (top right).
3. Choose Structured → upload file → enter a table name.
4. Wait for upload; table appears under Bronze in the left catalog.
5. Click the table to preview rows and types.

Add your first table

UI labels

The sidebar group is Data Warehouse; the product is Lake. Both go to /data.

Gold — publish for modeling (~5 min)

1. With Bronze (or Silver) open, click Publish Gold.
2. Name the artifact.
3. Select use case:
   • Machine learning — pick target column (100+ rows)
   • Forecast — pick time column (30+ rows, valid time series)
4. Validate → fix errors → Publish.
5. Confirm artifact under Gold tab, status ready.

Publish for modeling

Skip Silver when you can

One clean CSV? Bronze → Gold is enough. Use Lab or Silver when you need profiling, joins, or filters.

Train (~10–20 min)

Machine Learning

1. Sidebar → Machine Learning → New model.
2. Published dataset (Gold) → select your artifact.
3. Confirm target → name model → Train.
4. Training queue → open model when complete.

Forecast

1. Sidebar → Forecast → new build.
2. Select your Gold forecast dataset.
3. Set interval (daily, monthly, etc.) → Train.

Deploy (~5 min)

1. Open model/forecast detail (/ml/{id} or /forecast/{id}).
2. Generate API key — save immediately.
3. API Integration tab for endpoint and sample code.

ML deployment · Forecast deployment

Common mistakes

Mistake	What happens	Fix
Trying to upload CSV in ML wizard	No upload option (except samples)	Use Lake → Gold first
Gold missing in wizard	Publish failed or wrong use case	Data → Gold, status ready, ML vs Forecast flag
Validation failed	Row count or column types	ML needs 100+ rows; Forecast needs parseable dates
File too large	Upload error	Stay under 100 MB per file

Next steps

• ML build guide
• Forecast build guide
• Usage & monitoring
• User journeys — Lab path if data needs profiling

Validation failed at publish?

Fix column types or row counts in the publish dialog — Gold must be ready before it appears in the wizard.