Data Science Example: Landslide Analysis in Oregon and Washington

Landslide activity and impact are important issues in the Pacific Northwest.  Landslides worldwide cause economic losses of around $4B annually (EM-DAT 2007), with nationwide landslide costs around $1 to $2 billion in damages (USGS science features, June, 11, 2012).  In Washington state, annual emergency work on unstable slopes along highways costs between $200 and $250M (Kadri, 2015).  At the location of the Oso landslide near Oso, WA, financial losses are over $65M (Kadri, 2015), with 42 homes destroyed and indirect costs expected to exceed $100M.  Prior to the Oso landslide, Washington’s Aldercrest-Banyon slide (1999) was the costliest landslide in the state, with $110M in losses.  The US Geological Survey also estimates that deaths nationwide attributed to landslides range between 25 and 50 annually.

In the Python Jupyter notebook below, landslide data from oregon is extracted, analyzed, and run thru a series of analysis techniques including decision trees/random forest analysis.