Crafte Sessions analysis 2026 09 29 2026 10 29

sessions_analysis-2026-09-29-2026-10-29.csv · 2,246 rows · 15 columns · definition v3 saved 2026-10-29 14:08

What do you want to find out?

Start with the question. It decides which steps you see, what the models aim at and what counts as a good result. You can add more analyses to this dataset later, each with its own question, and reuse the work between them.

New analysis

1. What do you want to find out?

Predict an eventWill this person or case have the outcome? (classification)e.g. Who is likely to order?Predict a quantityHow much, how many or how long? (regression)e.g. How much will they spend?Find the biggest leverWhat could we change that would move the outcome most?e.g. What would raise the order rate?Describe groupsWhat kinds of people or cases are there, and what sets them apart?e.g. What types of visitor do we have?Test or estimateIs a difference real? What distribution does a value follow?e.g. Is tablet really worse at checkout?

2. Which column is the outcome?

For "test or estimate" this is optional: the value to fit or compare. Problem type Proposed from the column's values; change it if it's wrong.

3. When would the prediction be made?

Use segmentationCompare a model on the target segments with one trained on everything.

Name and purpose

Name What decision will this help make? (optional)

Saving makes this the analysis you're working on. Feature sets, segments and models stay with the dataset, so other analyses can reuse them.