10

It's old question. However, I could not find satisfactory answer. Decided to post this question. I have a .csv file with column heading. How can I import this .csv file into Mathematica assigning column names as list names in short and sweet way. For example, I can use the following codes.

mydata = Import[file, Path -> path];
names = Flatten[mydata[[;; 1]]]; 
Do[listnames[names[[i]]] = 
                     mydata[[2 ;;]][[All, i]], {i, 1, Dimensions[mydata[[2 ;;]]][[2]]}];

But I need to write listnames["Col names in .csv file"] to refer to each list. I can not refer .csv column names as list names. I need to figure out this because I have over 1000 column headings in my .csv data. Please help me to figure out his without long codes.

m_goldberg
  • 107,779
  • 16
  • 103
  • 257
ramesh
  • 2,309
  • 16
  • 29

1 Answers1

16

Update

Inspired by Andy Ross

header = First @ iris;

header = StringReplace[header,"."->""](Because Dot is protected in Mathematica)

data = Rest @ iris;

assign[name_, value_] := Evaluate[ToExpression[name]] = value;

Thread[ assign[header, Transpose@data] ]

====================================================================

If you are using Mathematica 10, there is a new data type called Dataset. Take the iris data for example. Suppose the iris.csv is in d:\.

iris = Import["d:/iris.csv","CSV"];

header = iris[[1]];


data = iris[[2;;]];


irisDataset = Thread[header -> #]& /@ data // Map[Association] // Dataset

and it will display like this:

enter image description here

Please refer to Dataset reference for more information.

Community
  • 1
PhilChang
  • 598
  • 4
  • 11
  • Thank you for your time. Yes, I am using Mathematica 10. I tried your codes but it does not give me the results I desire. I want the column names of .csv file as list names so that I can refer each .csv column data by their column name as list name in Mathematica. In your example, I should be able to refer each column as a separate list with its name being column heading. I should be able to refer first column as a list as Sepal.Length, second column as another list as Sepal.Width and so on.... – ramesh Aug 12 '14 at 00:54
  • @rkadhikari I advise reading the documentation page for Dataset -- it's pretty trivial to take a dataset and extract the raw columns from it, but you might find the Dataset itself much easier to work with for some kinds of things. – Taliesin Beynon Aug 12 '14 at 02:33
  • Thanks for your time, PhilChang. That's what I was looking for. You are really genius!!:) – ramesh Aug 12 '14 at 02:34
  • 4
    @PhilChang SemanticImport makes it really easy to import a tabular file as a Dataset if they have a reasonable structure -- no need to parse it to and format it yourself. So for example your code could just be replaced with SemanticImport["https://raw.githubusercontent.com/pydata/pandas/master/pandas/tests/data/iris.csv"] – Taliesin Beynon Aug 12 '14 at 02:36
  • @Taliesin, Thanks for your advice. I will follow your suggestion. I am new to Mathematica, started to like because of friends like you :) – ramesh Aug 12 '14 at 02:40
  • @TaliesinBeynon,Thank you, SemanticImport is really powerful and amazing! – PhilChang Aug 12 '14 at 02:41
  • @rkadhikari,I feel so flattered, :). I am also new to Mathematica. I like R syntax more. you can fit a linear model just using mymodel = lm(y~x1+x2+x3,data = mydata). It is so intuitive. – PhilChang Aug 12 '14 at 02:46
  • 2
    An alternative way to create a Dataset is irisDataset = Dataset[Inner[Rule, header, Transpose[data], Association]]. – Ernst Stelzer Aug 12 '14 at 07:38
  • @TaliesinBeynon We use Excel files (.XLSX) to import data as well as export data. It provides us with a simple interface that can be understood and used by most people. Giving columns distinct, complete and reasonable names means the order of the columns does not matter and provides a named rather than a numbered access to the data. In this sense Association and Dataset provide serious progress within Mathematica. In a further step, we also named all rows, which means each cell is identified by pairs of names. Please do not forget that Excel sheets are regularly sorted. – Ernst Stelzer Aug 12 '14 at 07:56
  • @ErnstStelzer, Excellent! you bring me a new understanding of Inner – PhilChang Aug 12 '14 at 09:20
  • Thank you all for your time to answer this question. I tried everyone's codes. It seems that I was not able to communicate my concern more clearly. I was not trying to create a Dataset or any data structure. I was just trying to assign .csv column names as list names referring to each column once we import the .csv file into Mathematica. Anyways, PhilChang's updated codes works perfectly fine for my purpose. Thank you guys again. – ramesh Aug 12 '14 at 14:53
  • @ErnstStelzer you get named rows by using an association of association instead of a list of them. – Rojo Aug 12 '14 at 18:05
  • The last expression is the use case for AssociationThread: irisDataset = AssociationThread[header -> #] & /@ data // Dataset – masterxilo May 17 '16 at 14:24
  • 3
    @PhilChang You can achieve the same result as in your answer with just Import with the Dataset representation and the HeaderLines option: Import["d:/iris.csv", {"CSV", "Dataset"}, "HeaderLines" -> 1] – barrywardell Nov 24 '17 at 00:42