Question

I have text file(test.data), which include some values and class name, for example

4.5,3.5,U1
4.5,10.5,U2
4.5,6,U1
3.5,10.5,U2
3.5,10.5,U2
5,7,U1
7,6.5,U1

I need to classify this data to rows(matrix), where 1,2 row etc is data, last one is class. So I started with this code:

reader = csv.reader(open('test.data', 'r'))
result = []
for row in reader:
    result.append(row)
print result

output:

[['4.5', '3.5', 'U1'], ['4.5', '10.5', 'U2'], ['4.5', '6', 'U1'], ['3.5', '10.5', 'U2'], ['3.5', '10.5', 'U2'], ['5', '7', 'U1'], ['7', '6.5', 'U1']]

This all work ok, but now I need from this data make matrix classification. In this case I want to make matrix:

test data=[data1, data2,.....,class name1]
test data2=[data1, data2,.....,class name2]...

I need this “matrix”(test data,test data2), because I will then choose from every test data just 2/3 data which will be named “choosen“, other 1/3 data must stay in test data,….

So what I need as output:

choosen=[data,data,......class name1]         # 2/3 from every **test.data**
test data=[data1, data2,.....,class name1]    # other 1/3 from test data
test data2=[data1, data2,.....,class name1]   # other 1/3 from test data 2

.

. . . .

.

Many thanks for help

---

EDIT2:

If I use your code I get:

{
    'U1': [
              ['4.5','3.5'],
              ['4.5','6'],
              ['5','7'],
              ['7','6.5']
          ],
    'U2': [
              ['4.5','10.5'],
              ['3.5','10.5'],
              ['3.5','10.5']
          ]
}

But I don’t have everytime this data:

4.5,3.5,U1
4.5,10.5,U2
4.5,6,U1
3.5,10.5,U2
3.5,10.5,U2
5,7,U1
7,6.5,U1

I have also:

4.5,3.5,4.5,10.5, U1
3.5,10.5,3.5,10.5,U2
4.5,12.5,3.5,12.5,U2

……
(so I don’t know that class is on second row as you write on your code), but I know that last row is CLASS

So how can I change your code:

reader = csv.reader(open('test.data', 'r'))
result = {}
for row in reader:
    uclass=row[2]                      #-------> must be last row not second !!!!
    if result.has_key(uclass):
        result[uclass].append([row[0],row[1]])   #---->not just 2 row's, on other data I have for example 5 rows..
    else:
        result[uclass]=[[row[0],row[1]]]        #---->not just 2 row's, on other data I have for example 5 rows..
print repr(result)

Answer 1

Edit: Original code snippet modified to handle N-column input (last is class). This requirement was mentioned by the OP in a later reply.

reader = csv.reader(open('test.data', 'r'))
result = {}
for row in reader:
    uclass=row[-1]
    if result.has_key(uclass):
        result[uclass].append(row[0:-1])
    else:
        result[uclass]=row[0:-1]
print repr(result)

I’m still not completely sure about the second half of your problem.
Something is missing in how you explained it perhaps.

If I understand this correctly you want a different list for each class?

If this is the case a dictionary should do what you want:

reader = csv.reader(open('test.data', 'r'))
result = {}
for row in reader:
    uclass=row[2]
    if result.has_key(uclass):
        result[uclass].append([row[0],row[1]])
    else:
        result[uclass]=[[row[0],row[1]]]
print repr(result)

result will look like:

{
    'U1': [
              ['4.5','3.5'],
              ['4.5','6'],
              ['5','7'],
              ['7','6.5']
          ],
    'U2': [
              ['4.5','10.5'],
              ['3.5','10.5'],
              ['3.5','10.5']
          ]
}

For the data skipping you can use the list slice skip option available in newer Pythons:

someList[<start>:<end>:<skip>]

so on a list like:

someList=[0,1,2,3,4,5,6,7,8,9]

using skip sliceing such as:

someList[0:10:2]

gives:

[0,2,4,6,8]