Problem with IJulia notebook and GLM.jl

[tcfuji]

Currently using Julia 3.1, GLM.jl 0.4 and IJulia 1.15
The following works on the Julia and IJulia console:

In [3]: lm1 = lm(OptDen ~ Carb, form)
Out[3]:
DataFrameRegressionModel{LinearModel{DensePredQR{Float64}},Float64}:

Coefficients:
               Estimate  Std.Error  t value Pr(>|t|)
(Intercept)  0.00508571 0.00783368 0.649211   0.5516
Carb           0.876286  0.0135345  64.7444    <1e-6

However, on IJulia notebook, I get the following output:

In [6]: lm1 = lm(OptDen ~ Carb, form)

`convert` has no method matching convert(::Type{Array{Symbol,1}}, ::Array{Any,1})
while loading In[6], in expression starting on line 1

 in convert at base.jl:13
 in ModelFrame at ~/.julia/DataFrames/src/statsmodels/formula.jl:238
 in fit at ~/.julia/DataFrames/src/statsmodels/statsmodel.jl:52
 in lm at ~/.julia/GLM/src/lm.jl:43

[stevengj]

Can you type versioninfo() in the console and in the notebook? Maybe you are not using the same Julia version in the two cases (in which case you can fix it by typing Pkg.build("IJulia") in the console and re-launching the notebook).

[tcfuji]

I executed versioninfo() on both the console and notebook. Both produced the same output. I even executed Pkg.build("IJulia"), but lm1 = lm(OptDen ~ Carb, form) still won't give the correct output on the notebook.

[stevengj]

@TFGIT, can you post a self-contained, complete code sample that exhibits your problem? @simonster, @IainNZ, can you reproduce this behavior?

[IainNZ]

using RDatasets
using GLM
form = dataset("datasets", "Formaldehyde")
lm1 = lm(OptDen ~ Carb, form)

Doesn't produce any errors for me in an IJulia notebook. Julia 0.3.0, IJulia 0.1.15, OSX 10.9

[tcfuji]

This has become more complicated...

Actually I can execute

using RDatasets
using GLM
form = dataset("datasets", "Formaldehyde")
lm1 = lm(OptDen ~ Carb, form)

in that sequence alone. However, mixed in with some other commands, I am still getting the same output. The IJulia notebook is included here: http://nbviewer.ipython.org/github/tfgit/statlearning-with-julia/blob/master/src/3.%20Linear%20Regression.ipynb

I am also getting the same output in JuliaBox: http://nbviewer.ipython.org/github/tfgit/statlearning-with-julia/blob/master/src/3.%20Linear%20Regression%20JuliaBox%20version.ipynb

My guess is that some of the prior commands may be to blame.

[simonster]

I can reproduce on 0.3 from the REPL with:

julia> using Color

julia> using RDatasets

julia> using GLM
Warning: could not import Base.add! into NumericExtensions

julia> form = dataset("datasets", "Formaldehyde")
6x2 DataFrame
| Row | Carb | OptDen |
|-----|------|--------|
| 1   | 0.1  | 0.086  |
| 2   | 0.3  | 0.269  |
| 3   | 0.5  | 0.446  |
| 4   | 0.6  | 0.538  |
| 5   | 0.7  | 0.626  |
| 6   | 0.9  | 0.782  |

julia> lm1 = lm(OptDen ~ Carb, form)
ERROR: `convert` has no method matching convert(::Type{Array{Symbol,1}}, ::Array{Any,1})
 in convert at base.jl:13
 in ModelFrame at /Users/simon/.julia/v0.3/DataFrames/src/statsmodels/formula.jl:238
 in fit at /Users/simon/.julia/v0.3/DataFrames/src/statsmodels/statsmodel.jl:52
 in lm at /Users/simon/.julia/v0.3/GLM/src/lm.jl:43

Evidently not an IJulia issue. Probably related to JuliaLang/julia#8631

[tcfuji]

You're right. Sorry for the trouble.