Linear model from test

src/extractor/AmrexCastro.jl

 
 include("../helper/helperSystem.jl")
 
-@time import DataFrames
+import DataFrames
+import GLM
 
 struct AmrexCastro <: AbstractAmrex
     app::String
@@ -23,7 +24,10 @@ struct AmrexCastro <: AbstractAmrex
             "castro.max_grid_size",
         ],
         Dict{String,Array{String}}(
-            "plots_size" => ["max_step", "amr.plot_int", "amr.n_cell", "amr.max_level"],
+            "plots_size" => ["amr.nplot_files", "amr.ncells", "amr.max_level"
+                             ],
+            "checks_size" => ["amr.check_int", "amr.ncheck_files", "amr.ncells", "amr.max_level",
+                             "amr.regrid_int"],
         ),
         runlogFile,
     )
@@ -93,9 +97,10 @@ end
 
 function _run_linear_model_plots_size(extractor::AmrexCastro, X::DataFrames.DataFrame)
 
-    XColumns::Array{String} = extractor.outputs["plots_size"]
-
     outputDirs = helper_get_prefix_directories(extractor.outputPrefix)
+  
+    # single column DataFrame Y in the linear model
+    plotsSizesData = DataFrames.DataFrame(plots_sizes = Int64[]) 
 
     for outputDir in outputDirs
 
@@ -111,19 +116,34 @@ function _run_linear_model_plots_size(extractor::AmrexCastro, X::DataFrames.Data
         for plotFileDir in plotFileDirs
             sizeDir::Int64 = helper_get_directory_size(plotFileDir)
             plotsSize += sizeDir
-            println(plotFileDir, " size: ", sizeDir)
         end
-
-        println("plots_size: ", plotsSize)
-
-        # println(plotFileDirs)
-
-        # todo add nprocs to the X matrix
-        # extractor.runlogFile
-
+        
+        push!(plotsSizesData, [plotsSize])
     end
-
-
-
-
+   
+    # Prepare the X independent variables in the linear model
+    XNames::Array{String} = extractor.outputs["plots_size"]
+      
+    for XName in XNames
+       columnName = Symbol(XName)
+       # this syntax [!, symbol] does not make a copy, use [:, symbol] for copies
+       plotsSizesData[!,columnName] = X[!,columnName]
+    end
+    
+    # formula from https://discourse.julialang.org/t/glm-jl-with-unknown-column-names/20692/5
+    response = Symbol(names(plotsSizesData)[1])
+    println(response)
+    
+    predictors = Symbol.(names(plotsSizesData)[2:end])
+    println(predictors)
+    
+    println(plotsSizesData)
+    f = @eval(GLM.@formula($response ~ (+)(1, $(predictors...))))
+    println(f)
+    
+    ols = GLM.lm(f, plotsSizesData)
+    println(ols)                            
+    
+    #import Plots
+    #display(Plots.plot(X, Y))
 end

src/helper/helperSystem.jl

@@ -38,6 +38,7 @@ function helper_get_prefix_files(prefix::String, path::String)::Array{String}
 
     pattern::String = string(prefix, "*")
     files = Glob.glob(pattern, path)
+    return files
 end
 
 

test/test_AmrexCastro.jl

@@ -77,13 +77,3 @@ end
 
 test_AmrexCastro()
 
-X = [1, 2, 3, 4, 5, 6]
-Y = [2, 4, 7, 9, 11, 13]
-
-#@time import GLM
-
-#ols = GLM.lm(GLM.@formula(Y ~ X), data)
-# println(ols)
-
-@time import Plots
-display(Plots.plot(X, Y))