Merge pull request #130 from dwijenchawra/master

Fix for #121

Author: mykelk

Project.toml

@@ -26,7 +26,7 @@ Documenter = "e30172f5-a6a5-5a46-863b-614d45cd2de4"
 
 
 [compat]
-DataFrames = "0.21,0.22"
+DataFrames = "0.22,1.0,1.1"
 DataStructures = "0.11,0.12,0.13,0.14,0.15,0.16,0.17,0.18"
 Discretizers = "3.0"
 Distributions = "0.17,0.18,0.19,0.20,0.21,0.22,0.23,0.24"

src/DiscreteBayesNet/discrete_bayes_net.jl

@@ -244,7 +244,7 @@ function statistics(dag::DAG, data::DataFrame)
 
     parents = [inneighbors(dag, i) for i in 1:n]
     ncategories = [Int(infer_number_of_instantiations(data[!,i])) for i in 1 : n]
-    datamat = convert(Matrix{Int}, data)'
+    datamat = Matrix{Int}(data)'
 
     statistics(parents, ncategories, datamat)
 end
@@ -255,8 +255,7 @@ function statistics(bn::DiscreteBayesNet, target::NodeName, data::DataFrame)
     targetind = bn.name_to_index[target]
     parents = inneighbors(bn.dag, targetind)
     ncategories = [Int(infer_number_of_instantiations(data[!,i])) for i in 1 : n]
-    datamat = convert(Matrix{Int}, data)'
-
+    datamat = Matrix{Int}(data)'
     statistics(targetind, parents, ncategories, datamat)
 end
 

src/DiscreteBayesNet/greedy_hill_climbing.jl

@@ -65,7 +65,7 @@ function Distributions.fit(::Type{DiscreteBayesNet}, data::DataFrame, params::Gr
 
     n = ncol(data)
     parent_list = map!(i->Int[], Array{Vector{Int}}(undef, n), 1:n)
-    datamat = convert(Matrix{Int}, data)'
+    datamat = Matrix{Int}(data)'
     score_components = bayesian_score_components(parent_list, ncategories, datamat, params.prior, params.cache)
 
     while true

src/DiscreteBayesNet/greedy_thick_thinning.jl

@@ -20,7 +20,7 @@ function Distributions.fit(::Type{DiscreteBayesNet}, data::DataFrame, params::Gr
 
     n = ncol(data)
     parent_list = map!(i->Int[], Array{Vector{Int}}(n), 1:n)
-    datamat = convert(Matrix{Int}, data)'
+    datamat = Matrix{Int}(data)'
     score_components = bayesian_score_components(parent_list, ncategories, datamat, params.prior, params.cache)
 
     while true

src/DiscreteBayesNet/scan_greedy_hill_climbing.jl

@@ -87,7 +87,7 @@ function Distributions.fit(::Type{DiscreteBayesNet}, data::DataFrame, params::Sc
 
     n = ncol(data)
     parent_list = map!(i->Int[], Array{Vector{Int}}(undef, n), 1:n)
-    datamat = convert(Matrix{Int}, data)'
+    datamat = Matrix{Int}(data)'
     score_components = bayesian_score_components(parent_list, ncategories, datamat, params.prior, params.cache)
 
     # 0 depth

src/DiscreteBayesNet/structure_scoring.jl

@@ -111,7 +111,7 @@ function bayesian_score(bn::DiscreteBayesNet, data::DataFrame, prior::DirichletP
     n = length(bn)
     parent_list = Array{Vector{Int}}(undef, n)
     ncategories = Array{Int}(undef, n)
-    datamat = convert(Matrix{Int}, data)'
+    datamat = Matrix{Int}(data)'
 
     for (i,cpd) in enumerate(bn.cpds)
         parent_list[i] = inneighbors(bn.dag, i)
@@ -168,7 +168,7 @@ function bayesian_score_components(bn::DiscreteBayesNet, data::DataFrame, prior:
     n = length(bn)
     parent_list = Array{Vector{Int}}(undef, n)
     ncategories = Array{Int}(undef, n)
-    datamat = convert(Matrix{Int}, data)'
+    datamat = Matrix{Int}(data)'
 
     for (i,cpd) in enumerate(bn.cpds)
         parent_list[i] = inneighbors(bn.dag, i)

src/Factors/factors_dataframes.jl

@@ -7,7 +7,7 @@
 Convert a Factor to a DataFrame
 """
 function Base.convert(::Type{DataFrame}, ϕ::Factor)
-    df = DataFrames.DataFrame(pattern(ϕ))
+    df = DataFrames.DataFrame(pattern(ϕ), :auto)
     DataFrames.rename!(df, [f => t for (f, t) = zip(names(df), names(ϕ))] )
     df[!,:potential] = ϕ.potential[:]
 

src/gibbs.jl

@@ -241,7 +241,7 @@ function gibbs_sample_main_loop(
 
     end
 
-    return convert(DataFrame, t), (now() - start_time).value
+    return DataFrame(t), (now() - start_time).value
 end
 
 """

src/sampling.jl

@@ -169,7 +169,8 @@ function get_weighted_dataframe(bn::BayesNet, nsamples::Integer, evidence::Assig
     end
     t[:p] = w / sum(w)
 
-    convert(DataFrame, t)
+
+    t = DataFrame(t)
 end
 
 get_weighted_dataframe(bn::BayesNet, nsamples::Integer, pair::Pair{NodeName}...) =

test/test_discrete_bayes_nets.jl

@@ -30,7 +30,7 @@ let
 	n = length(bn)
 	parent_list = Array{Vector{Int}}(undef, n)
 	bincounts = Array{Int}(undef, n)
-	datamat = convert(Matrix{Int}, data)'
+	datamat = Matrix{Int}(data)'
 
 	for (i,cpd) in enumerate(bn.cpds)
 		parent_list[i] = inneighbors(bn.dag, i)