PanelScope
What it does
It helps identify the largest set of variables that can be analysed together while preserving a user-defined proportion of complete panel observations. Given a country-year, firm-year, bank-year, or other entity-year dataset, PanelScope can measures how the complete-case sample changes as variables are combined, and reports the variable set that retains the most usable data subject to a stated retention requirement. It is intended as a preparatory step before estimation, making the trade-off between variable breadth and sample retention explicit and reproducible.
Why I built it
I struggled finding the best selection of variables for a dataset of 12 features. Then my professor sent an 18,000 feature dataset.
Status
Working prototype. Not maintained. Fork it rather than depend on it.