Data prep · Beginner · 10 min
Configuring Columns Correctly
Data preparation
The single highest-leverage skill: teaching the platform what each column means.
Steps
Follow along
Each step is something you can do in the platform right now.
01
Step 1
Upload a dataset and open the column-review view in Data Synthesis.
02
Step 2
Verify every detected type: continuous, categorical, discrete or boolean.
03
Step 3
Mark identifier columns (sample IDs, dates, notes) so they are excluded from modeling.
04
Step 4
Designate the outcome column you want to predict later as the target.
05
Step 5
Check that percentage columns are numeric, not text (fix "0,5" style formatting first).
06
Step 6
Save the configuration and run a quick data synthesis to confirm the types stick.
Notes
Pro tips
What experienced users wish they knew the first time.
A mislabeled column type is the most common cause of weak results — review types once, save yourself hours later.
Columns with mixed text and numbers get classified as categorical. Clean them before upload.
Want the details? This tutorial pairs with the docs page data prep · column roles.
Read the docs →