Hello, This question has already been asked several times (here for example How to set a index in Julia's DataFrame? - Stack Overflow ) but I still don’t understand why DataFrame.jl doesn’t have an index_col parameter like there is in Pandas ( pandas.read_csv — pandas 2.0.3 documentation ). Is it beca…

I think that OP wasn’t asking about the most efficient way, but the simplest approach. If you want to keep your code simple, my recommendation is the following: df = DataFrame(id = 1:5, weight= 56:60, age = 26:30) df[df.id .== 1, :age] # for returning a value/vector df[df.id .== 1, [:age]] # fo…

You can simply right df[df.patient_id .== 1234, :age]

I have wanted a simpler way to do this in the past too. The short versions are somewhat hard to discover on your own. ==(value) is a shorthand for x -> x == value. . syntax retrieves the column you are interested in from the filtered data frame. (df.age is equivalent to df[!, :age].) only checks …

That would allocate a vector to make the comparison. I would consider (for nice, readable syntax): patient_row(df, id) = findfirst(r -> r.patient_id == id, eachrow(df)) # or `only(findall())` if one wants to ensure that the id is unique # produes a DataFrameRow if `cols` is Vector get_patient(df,…

This will always return a vector: [image] alfaromartino: df[df.id .== 1, :age] # for returning a value/vector julia> df = DataFrame(id = 1:5, weight= 56:60, age = 26:30); julia> df[df.id .== 3, :age] 1-element Vector{Int64}: 28 To return a value, you would need to do one of the following: …

Indeed row-lookup is an important use case. And as it was already commented here it is not easy to design a good API. The biggest issue is that the condition you might want to use could return exactly one row, or multiple rows (where 0 rows is a special case of multiple). I want to add something mo…

An important use case I often need: if a table has an index (key column), this defines a mapping between it and any other column, which can then be applied to other tables to add columns according to those mappings. A simple and efficient API for that would be nice (if it does not exist already).

What you describe is handled by joins (if I understand your need correctly).

Joins can handle this but quite slowly in my experience. Instead, I normally use: lookup = Dict(df1.key_col .=> df1.value_col) df2.values = [lookup[key] for key in df2.some_col] But I guess one could argue that I should not be using a DataFrame, but a TypedTables.DictTable.

Thank you all for your answers. The solution df[df.patient_id .== 1234, :age] |> only seems to be the simplest !

Ok, no problem, I’ll go and comment on the git discussion. In fact, I’m in a fairly simple situation where each line is unique (an id doesn’t appear twice).

Why is it so complicated to access a row in a DataFrame?

General Usage

Topic		Replies	Views
Any plan for functionality like Pandas loc? General Usage dataframes	15	1269	May 27, 2022
Easier access to DataFrame's elements General Usage	6	1851	September 28, 2018
How to select dataframe rows in Julia 1.6? New to Julia question	1	5372	May 7, 2021
Is there a faster way to extract single rows of a dataframe than the filter function General Usage	3	685	February 18, 2020
Data structure for convenient access to tabular data General Usage dataframes	5	452	February 13, 2023

Why is it so complicated to access a row in a DataFrame?

Related topics