1 Data
No data handy? — 120 runs of a fictional reactor with a known answer. You can also download it to open in Excel; its first sheet walks through what to do and what to expect, and the last sheet gives the equations the data came from.
Assign columns
Mark the quantity you want to predict as Target, the variables that explain it as Feature, and everything else as Ignore.
Preview data
2 Candidate functions
Choose which families of terms the search may draw from. More families means a richer search but a slower one, and raises the chance of a term fitting noise.
0 candidate terms
Search settings
3 Ranked models
Selected model
Cross-validation
4 Export
The Python script reproduces this exact fit with pip install jaxsr. Download the data
alongside it so the script has something to read.