Skip to content

Commit

Permalink
Update xlsout.jl
Browse files Browse the repository at this point in the history
  • Loading branch information
@mwsohn
mwsohn committed Jun 12, 2024
1 parent 26ef0bc commit d2ec84d
Showing 1 changed file with 47 additions and 108 deletions.
155 changes: 47 additions & 108 deletions src/xlsout.jl
View file
Original file line number Diff line number Diff line change
Expand Up @@ -35,17 +35,18 @@ function hltest(glmout,q = 10)
return (hlstat, dof, pval)
end

function nagelkerke(glmout)
return StatsBase.r2(glmout,:Nagelkerke)
end
# For these functions, use r2(GLM, :MacFadden) or r2(GLM,:Nagelkerke)
# function nagelkerke(glmout)
# return StatsBase.r2(glmout,:Nagelkerke)
# end

function macfadden(glmout)
return StatsBase.r2(glmout,:McFadden)
end
# function macfadden(glmout)
# return StatsBase.r2(glmout,:McFadden)
# end


"""
glmxls(glmout::DataFrames.DataFrameRegressionModel, workbook::PyObject, worksheet::AbstractString; labels::Union{Nothing,Label}=nothing,eform=false,ci=true, row = 0, col =0)
glmxls(glmout::DataFrames.DataFrameRegressionModel, workbook::PyObject, worksheet::AbstractString; eform=false,ci=true, row = 0, col =0)
Outputs a GLM regression table to an excel spreadsheet.
To use this function, `PyCall` is required with a working version python and
Expand All @@ -55,7 +56,6 @@ or a value of a variable in a `Label`, the label will be output. Options are:
- `glmout`: returned value from a GLM regression model
- `workbook`: a returned value from xlsxwriter.Workbook() function (see an example below)
- `worksheet`: a string for the worksheet name
- `labels`: an option to specify a `label` dictionary (see an example below)
- `eform`: use `eform = true` to get exponentiated estimates, standard errors, or 95% confidence intervals
- `ci`: use `ci = true` (default) to get 95% confidence intervals. `ci = false` will produce standard errors and Z values instead.
- `row`: specify the row of the workbook to start the output table (default = 0 (for row 1))
Expand All @@ -76,7 +76,7 @@ PyObject <xlsxwriter.workbook.Workbook object at 0x000000002A628E80>
julia> glmxls(ols1,wb,"OLS1",labels = label)
julia> bivairatexls(df,:incomecat,[:age,:race,:male,:bmicat],wb,"Bivariate",labels = label)
julia> bivairatexls(df,:incomecat,[:age,:race,:male,:bmicat],wb,"Bivariate")
Julia> wb.close()
```
Expand All @@ -86,15 +86,14 @@ Alternatively, one can create a spreadsheet file directly. `PyCall` or `@pyimpor
does not need to be called before the function.
```
julia> glmxls(ols1,"test_workbook.xlsx","OLS1",labels = label)
julia> glmxls(ols1,"test_workbook.xlsx","OLS1")
```
"""
function glmxls(glmout,wbook::PyObject,wsheet::AbstractString;
labels::Union{Nothing,Label} = nothing,
eform::Bool = false, ci = true, row = 0, col = 0)
function glmxls(glmout,wbook::PyObject,wsheet::AbstractString; labels = nothing,
eform::Bool = false, ci = true, row = 0, col = 0, robust::Symbol = nothing)

if (typeof(glmout) <: StatsModels.RegressionModel) == false
if (typeof(glmout) <: StatsModels.TableRegressionModel) == false
error("This is not a regression model output.")
end

Expand Down Expand Up @@ -162,13 +161,13 @@ function glmxls(glmout,wbook::PyObject,wsheet::AbstractString;
vals[i] = ""
end

# use labels if exist
# column labels
if labels != nothing
if vals[i] != ""
valn = vals[i] == "true" ? 1 : parse(Int,vals[i])
vals[i] = vallab(labels,Symbol(varname[i]),valn)
if haskey(labels,Symbol(varname[i]))
varname[i] = labels[Symbol(varname[i])]
else
varname[i] = string(varname[i])
end
varname[i] = varlab(labels,Symbol(varname[i]))
end
end
for i = 1:nrows
Expand Down Expand Up @@ -220,6 +219,10 @@ function glmxls(glmout,wbook::PyObject,wsheet::AbstractString;
t.write(r,c+1,tdata.cols[1][i],formats[:or_fmt])
end

# standard errors
if robust != nothing
end

if ci == true

if eform == true
Expand Down Expand Up @@ -265,17 +268,17 @@ function glmxls(glmout,wbook::PyObject,wsheet::AbstractString;
t.merge_range(r,c+1,r,c+4,nobs(glmout),formats[:n_fmt_center])

# degress of freedom
# r += 1
# t.write(r,c,"DF",formats[:model_name])
# t.merge_range(r,c+1,r,c+4,dof(glmout),formats[:n_fmt_center])
r += 1
t.write(r,c,"DF",formats[:model_name])
t.merge_range(r,c+1,r,c+4,dof(glmout),formats[:n_fmt_center])

# R² or pseudo R²
r += 1
if isa(linkfun,LogitLink)
t.write(r,c,"Pseudo R² (MacFadden)",formats[:model_name])
t.merge_range(r,c+1,r,c+4,macfadden(glmout),formats[:p_fmt_center])
t.merge_range(r,c+1,r,c+4,r2(glmout,:MacFadden),formats[:p_fmt_center])
t.write(r+1,c,"Pseudo R² (Nagelkerke)",formats[:model_name])
t.merge_range(r+1,c+1,r+1,c+4,nagelkerke(glmout),formats[:p_fmt_center])
t.merge_range(r+1,c+1,r+1,c+4,r2(GLM,:Nagelkerke),formats[:p_fmt_center])

# -2 log-likelihood
t.write(r+2,c,"-2 Log-Likelihood",formats[:model_name])
Expand Down Expand Up @@ -317,15 +320,16 @@ end
function glmxls(glmout,
wbook::AbstractString,
wsheet::AbstractString;
labels::Union{Nothing,Label} = nothing,
labels = nothing,
eform::Bool = false,
ci = true,
row = 0,
col = 0)
col = 0,
robust = nothing)

xlsxwriter = pyimport("xlsxwriter")

glmxls(glmout,xlsxwriter.Workbook(wbook),wsheet,labels=labels,eform=eform,ci=ci,row=row,col=col)
glmxls(glmout,xlsxwriter.Workbook(wbook),wsheet,labels=labels,eform=eform,ci=ci,row=row,col=col,robust=robust)
end


Expand Down Expand Up @@ -383,17 +387,11 @@ function bivariatexls(df::DataFrame,
wbook::PyObject,
wsheet::AbstractString;
wt::Symbol = nothing,
labels::Label = nothing,
row::Int = 0,
col::Int = 0,
column_percent::Bool = true,
verbose::Bool = false)

# labels
# if labels == nothing && "Labels" in metadatakeys(df)
# labels = load_labels(df)
# end

# colvar has to be a CategoricalArray and must have 2 or more categories
if isa(df[!,colvar], CategoricalArray) == false || length(levels(df[!,colvar])) < 2
error("`",colvar,"` is not a CategoricalArray or does not have two or more levels")
Expand Down Expand Up @@ -427,41 +425,25 @@ function bivariatexls(df::DataFrame,
t.set_column(c,c,40)
t.set_column(c+1,c+(nlev+1)*2+1,9)

# create heading
# create header
# column variable name
# It uses three rows
t.merge_range(r,c,r+2,c,"Variable",formats[:heading])

# 1st row = variable name
colvname = varlab(df,colvar)
t.merge_range(r,c+1,r,c+(nlev+1)*2+1,colvname,formats[:heading])
# header 1st row = variable name
t.merge_range(r,c+1,r,c+(nlev+1)*2+1,label(df,colvar),formats[:heading])

# 2nd and 3rd rows
# header 2nd and 3rd rows
r += 1

t.merge_range(r,1,r,2,"All",formats[:heading])
t.write_string(r+1,1,"N",formats[:n_fmt_right])
t.write_string(r+1,2,"(%)",formats[:pct_fmt_parens])

c = 3

# value labels for colvar
vals = vallab(df,colvar)

#
c += 3
for i = 1:nlev

# value label
# vals = string(colnms[i])
# if labels != nothing
# vals = value_label(df,colvar,colnms[i])
# end
# println(colvar, " ", colnms[i], " ", vals[colnms[i]])
if vals != nothing && haskey(vals,colnms[i])
vlab = vals[colnms[i]]
else
vlab = string(colnms[i])
end
t.merge_range(r,c+(i-1)*2,r,c+(i-1)*2+1,vlab,formats[:heading])
t.merge_range(r,c+(i-1)*2,r,c+(i-1)*2+1,colnms[i],formats[:heading])
t.write_string(r+1,c+(i-1)*2,"N",formats[:n_fmt_right])
t.write_string(r+1,c+(i-1)*2+1,"(%)",formats[:pct_fmt_parens])
end
Expand All @@ -470,7 +452,7 @@ function bivariatexls(df::DataFrame,
t.merge_range(r,c+nlev*2,r+1,c+nlev*2,"P-Value",formats[:heading])

# total
c = 0
c = col
r += 2
t.write_string(r,c,"All, n (Row %)",formats[:model_name])
if wt == nothing
Expand All @@ -488,23 +470,16 @@ function bivariatexls(df::DataFrame,
t.write(r,c+(nlev+1)*2+1,"",formats[:empty_border])

# covariates
c = 0
c = col
r += 1
for varname in rowvars

if verbose == true
println("Processing ",varname)
end

# sum of rowvars must be non-zero
# if sum(length(skipmissing(df[varname]))) == 0
# warn("`",varname,"` in `rowvars` is empty.")
# continue
# end

# print the variable name
# vars = string(varname)
vars = varlab(df,varname)
# variable name
vars = label(df,varname)

# determine if varname is categorical or continuous
if isa(df2[!,varname], CategoricalArray) || eltype(df2[!,varname]) == String
Expand Down Expand Up @@ -562,13 +537,7 @@ function bivariatexls(df::DataFrame,
r += 1
for i = 1:length(rowval)
# row value
# vals = string(rowval[i])

# if labels != nothing
# vals = vallab(labels,varname,rowval[i])
# end
vals = string(vallab(labels,varname,rowval[i]))
t.write_string(r,c,vals,formats[:varname_1indent])
t.write_string(r,c,rowval[i],formats[:varname_1indent])

# row total
t.write(r,c+1,rowtot[i],formats[:n_fmt_right])
Expand Down Expand Up @@ -601,7 +570,7 @@ function bivariatexls(df::DataFrame,
else
# continuous variable
df3=df2[completecases(df2[!,[varname]]),[varname,colvar]]
y = tabstat(df3, varname, colvar, table=false, labels=labels,wt=df3[wt])
y = tabstat(df3, varname, colvar, table=false, wt=df3[wt])

# variable name
t.write_string(r,c,string(vars,", mean (SD)"),formats[:model_name])
Expand Down Expand Up @@ -655,15 +624,14 @@ function bivariatexls(df::DataFrame,
rowvars::Vector{Symbol},
wbook::AbstractString,
wsheet::AbstractString;
labels::Union{Nothing,Label} = nothing,
row::Int = 0,
col::Int = 0)

xlsxwriter = pyimport("xlsxwriter")

wb = xlsxwriter.Workbook(wbook)

bivariatexls(df,colvar,rowvars,wb,wsheet,labels=labels,row=row,col=col)
bivariatexls(df,colvar,rowvars,wb,wsheet,row=row,col=col)

wb.close()
end
Expand Down Expand Up @@ -722,7 +690,6 @@ function univariatexls(df::DataFrame,
wbook::PyObject,
wsheet::AbstractString;
wt::Union{Nothing,Symbol} = nothing,
labels::Union{Nothing,Label}=nothing,
row = 0,
col = 0)

Expand All @@ -749,29 +716,6 @@ function univariatexls(df::DataFrame,
for i in 1:24
t.write_string(r+i,c,rownms[i],formats[:heading_left])
end
# t.write_string(r+2,c,"N Miss",formats[:heading_left])
# t.write_string(r+3,c,"N Used",formats[:heading_left])
# t.write_string(r+4,c,"Sum",formats[:heading_left])
# t.write_string(r+5,c,"Mean",formats[:heading_left])
# t.write_string(r+6,c,"SD",formats[:heading_left])
# t.write_string(r+7,c,"Variance",formats[:heading_left])
# t.write_string(r+8,c,"Minimum",formats[:heading_left])
# t.write_string(r+9,c,"P25",formats[:heading_left])
# t.write_string(r+10,c,"Median",formats[:heading_left])
# t.write_string(r+11,c,"P75",formats[:heading_left])
# t.write_string(r+12,c,"Maximum",formats[:heading_left])
# t.write_string(r+13,c,"Skewness",formats[:heading_left])
# t.write_string(r+14,c,"Kurtosis",formats[:heading_left])
# t.write_string(r+15,c,"Smallest",formats[:heading_left])
# t.write_string(r+16,c,"",formats[:heading_left])
# t.write_string(r+17,c,"",formats[:heading_left])
# t.write_string(r+18,c,"",formats[:heading_left])
# t.write_string(r+19,c,"",formats[:heading_left])
# t.write_string(r+20,c,"Largest",formats[:heading_left])
# t.write_string(r+21,c,"",formats[:heading_left])
# t.write_string(r+22,c,"",formats[:heading_left])
# t.write_string(r+23,c,"",formats[:heading_left])
# t.write_string(r+24,c,"",formats[:heading_left])

col = 1
for vsym in contvars
Expand All @@ -789,12 +733,8 @@ function univariatexls(df::DataFrame,
# non-missing values
len = size(df,1) - count(ismissing, df[!,vsym]) # sum(ismissing.(df[!,vsym]) .== false)

# if there is a label dictionary, pick up the variable label
varstr = string(vsym)
if labels != nothing
varstr = varlab(labels,vsym)
end

# pick up the variable label
varstr = label(df,vsym)
t.write_string(0,col,varstr,formats[:heading])
u = Stella.univariate(df[!,vsym]) #,wt=df[wt])
for j = 1:14
Expand Down Expand Up @@ -986,7 +926,6 @@ function dfxls(df::DataFrame,
wb.close()
end


function newfilename(filen::AbstractString)
while (isfile(filen))
# separate the name into three parts, basename (number).extension
Expand Down

0 comments on commit d2ec84d

Please sign in to comment.