Predictive Models Lab RNR/Geog 420/520 Spring
Predictive Models zImportant to understand what we are attempting to predict yThese models predict location yThis prediction is based on reasoned or measured relationships zNo predictive model is perfect zSome are more efficient than others
Broad Model Types zDeductive models are based on reasoning in which the conclusion follows necessarily the presented premises zInductive models base validity on observations about part of a class as evidence for a proposition about the whole class
Components of the Model zVariables --- DEM, relief, slope, shelter, ridge zStudy Group archaeological sites zControl Group random point locations zSuitable Statistical model --- logistic regression
Model Assumptions zThe Processes Ancient Humans Used to Select Site Locations Were Not Random zPart of the Site Selection Process Involved Selection for Favored Environmental Zones zConsequently, It Should be Possible to Identify Specific Environmental Signatures for Specific Groups of Archaeological Sites
Creating the Regression Models in GRID zSample values of variables at site and non- site locations zSubject SAMPLE results to regression in GRID (with some cautions) zResults of the regression include coefficients and a constant, or y-intercept zModel made by multiplying variables by coefficients --- sum of these variables is the model zModel then scaled between 0 and 1 to create a probability surface
Sample Dephdemhridge… …
Logistic Regression in GRID Grid: |> regression hsam.txt logistic brief <| coef # coef RMS Error = Chi-Square =
Making Probability Models X = X = X = X = X = X = X = Relief Shelter Dist. To Soil 1 Elevation Ridge Index Aspect Distance to Wadi Environmental Variables Regression Coefficients Weighted Variables Probability Model Sum of Weighted Variables Corrected Y-Intercept
Probability Models Group 1 Group 2
Model Strength