MATLAB: Parsing a text file

parsingtext file

Hello .I have the following question :I am reading a text file using the usual

textread(filename,'%s%s%s%f%f%d','delimiter',',').Sometimes though the file is corrupted and instead of a number as expected ,in one of the fields ,there is an INF .

How do I change the program so I can simply avoid these lines completely ?

Many thanks,

Constantin

Best Answer

Hi,

use readtable like suggested by Madhan. If it is always in column 8 of your data like your example given use logical indexing:

filename = 'filename.txt';
result = readtable(filename)
result(result.Var8==Inf,:) = []

using this example as file, where line 5 and 9 are corrupt:

EQ235464,Mining,Active,SPY,23.4,34,5,1,230
EQ235464,Mining,Active,SPY,23.4,34,5,2,230
EQ235464,Mining,Active,SPY,23.4,34,5,3,230
EQ235464,Mining,Active,SPY,23.4,34,5,4,230
EQ235464,Mining,Active,SPY,23.4,34,5,INF,230
EQ235464,Mining,Active,SPY,23.4,34,5,6,230
EQ235464,Mining,Active,SPY,23.4,34,5,7,230
EQ235464,Mining,Active,SPY,23.4,34,5,8,230
EQ235464,Mining,Active,SPY,23.4,34,5,INF,230
EQ235464,Mining,Active,SPY,23.4,34,5,10,230

results in:

result =
  10×9 table
       Var1         Var2        Var3      Var4     Var5    Var6    Var7    Var8    Var9
    __________    ________    ________    _____    ____    ____    ____    ____    ____
    'EQ235464'    'Mining'    'Active'    'SPY'    23.4     34      5        1     230 
    'EQ235464'    'Mining'    'Active'    'SPY'    23.4     34      5        2     230 
    'EQ235464'    'Mining'    'Active'    'SPY'    23.4     34      5        3     230 
    'EQ235464'    'Mining'    'Active'    'SPY'    23.4     34      5        4     230 
    'EQ235464'    'Mining'    'Active'    'SPY'    23.4     34      5      Inf     230 
    'EQ235464'    'Mining'    'Active'    'SPY'    23.4     34      5        6     230 
    'EQ235464'    'Mining'    'Active'    'SPY'    23.4     34      5        7     230 
    'EQ235464'    'Mining'    'Active'    'SPY'    23.4     34      5        8     230 
    'EQ235464'    'Mining'    'Active'    'SPY'    23.4     34      5      Inf     230 
    'EQ235464'    'Mining'    'Active'    'SPY'    23.4     34      5       10     230 
result =
  8×9 table
       Var1         Var2        Var3      Var4     Var5    Var6    Var7    Var8    Var9
    __________    ________    ________    _____    ____    ____    ____    ____    ____
    'EQ235464'    'Mining'    'Active'    'SPY'    23.4     34      5        1     230 
    'EQ235464'    'Mining'    'Active'    'SPY'    23.4     34      5        2     230 
    'EQ235464'    'Mining'    'Active'    'SPY'    23.4     34      5        3     230 
    'EQ235464'    'Mining'    'Active'    'SPY'    23.4     34      5        4     230 
    'EQ235464'    'Mining'    'Active'    'SPY'    23.4     34      5        6     230 
    'EQ235464'    'Mining'    'Active'    'SPY'    23.4     34      5        7     230 
    'EQ235464'    'Mining'    'Active'    'SPY'    23.4     34      5        8     230 
    'EQ235464'    'Mining'    'Active'    'SPY'    23.4     34      5       10     230

Best regards

Stephan

Related Solutions

MATLAB: Converting cell array from excel to numbers

Hi,

Before answering your question, know that most of the time the call of function find is useless. In your case you can do the following:

[numbers,strings,raws]=xlsread(FileName);
iHeight=strcmp(raws,'Height')
ShowRaws=raws(iHeight,3:end)
height=cell2mat(raws(iHeight,3:end))

As for your problem, it came from the "-" data which is transform into a '-' which canot be convert into number (even NaN).

I strongly suggest you you the function readtable:

data=readtable(FileName,'ReadVariableNames',false,'ReadRowNames' ,true,'TreatAsEmpty' ,'-');
height = data{'Height',2:end}

You can also do the following

height = data{1,2:end}

MATLAB: How to re-code numeric variables in a table to strings

JLC, you say "string", but it's very likely that you would be bettter off with categorical variables. Given you're original sample data ...

>> Var1 = [1; 2; 1; 2; 2; 1; 1];
>> Var2 = [2; 2; 1; 1; 1; 2; 1];
>> Var3 = [32.1; 34.6; 56.7; 65.3; 78.6; 23.4; 43.1];
>> t = table(Var1,Var2,Var3)
t =
  7×3 table
    Var1    Var2    Var3
    ____    ____    ____
     1       2      32.1
     2       2      34.6
     1       1      56.7
     2       1      65.3
     2       1      78.6
     1       2      23.4
     1       1      43.1
... here's what you probably want to do:
>> t.Var1 = categorical(t.Var1,1:2,{'Male' 'Female'});
>> t.Var2 = categorical(t.Var2,1:2,{'Old' 'Young'});
>> t
t =
  7×3 table
     Var1     Var2     Var3
    ______    _____    ____
    Male      Young    32.1
    Female    Young    34.6
    Male      Old      56.7
    Female    Old      65.3
    Female    Old      78.6
    Male      Young    23.4
    Male      Old      43.1

A few other things:

Don't name your table "Table". It will come back to bite you eventually.
No reason to concatenate [Var1,Var2,Var3] and then call array2table, which just pulls them apart. Call table directly.
Your "desired" result seems to have Var3 as a cell array, each cell containing a scalar double, You definitely don't want that.
It's possible to convert the entire table to a cell array and work on it and convert it back, as shown by ImageAnalyst. But you can work on the table directly as I showed above. The "Conversion to double from cell is not possible." errors you saw are because, as IA says, you were assigning text to only some elements of a double variable in the table. You need to assign to t.Var1 without subscripting. My code above shows that as part of converting to categorical.
If you really do want strings, do it like this:

genders = {'Male'; 'Female'}; t.Var1 = genders(t.Var1);
ages = {'Male'; 'Female'}; t.Var2 = ages(t.Var2);

Best Answer

Related Solutions

MATLAB: Converting cell array from excel to numbers

MATLAB: How to re-code numeric variables in a table to strings

Related Question