Common errors and problems
|Top Previous Next|
Common errors and problems
No matter how much care is taken with data preparation, it is not unusual to encounter problems when initially loading and running analyses on a data set. These are usually simple to rectify. Common error messages and their solutions are shown below.
1. Singular Matrix in Lower-Upper Decomposition routine.
Some data sets simply will not work with a Discriminant Analysis, if there is low (or zero) variability in the samples/sites, or if there are close correlations between sites or variables. The program will show two error messages; the first shows where in the data set the problem lies:
while the second states that no analysis is possible:
2. '' is not a valid floating point value.
This will occur if the raw data set holds blank columns or rows - ones that sum to zero. Remove blank columns and rows by using 'Handling zeros' in the Working Data tab. It may also occur if the raw data holds a blank cell. In some cases CAP will identify the problem cell which should be edited. Normally it is because the data has been prepared in a spreadsheet using blanks to represent zero values.
Occasionally, this error can occur because a blank space or a character has been accidentally entered into a cell outside the data matrix when it was being prepared in a spreadsheet program. To prevent this happening, it is good practice, before saving your data set as a .csv file, to highlight the first 10 or so blank rows and columns below and to the right of the data matrix, and press 'Delete'. This will clear the cells of any accidentally-entered contents.
3. I/O error 32 - access denied.
This will occur if the data file you are trying to open is currently being used by another program - normally the spreadsheet which was used to organise the data.
Close the file in other programs and try again.
4. When a data file is opened, all the data are dumped into the first cell, rather than opening in the grid properly.
This is usually because there are one or more blank cells in the second row of data. If the blanks are replaced by zeros, or another row with no blanks is put in the second position, the problem should disappear. Row 2 is the crucial one; blank cells are tolerated elsewhere.
5. I entered the data in a spreadsheet program, and saved it as a csv file. When I try to open it in CAP, the numbers are separated by ';' not ',' so the analysis will not run.
This is because some non-British systems use ; instead of , to separate values in a csv file. Open Windows Explorer, and change the file extension from csv to txt (i.e. Filename.csv becomes Filename.txt). Open the txt file in Word or another word-processing program. Use the Find/Replace function to replace every ; with a , then save the file again. Change the txt file extension back to csv. It should now open and run perfectly in CAP.
6. I attempted to run TWINSPAN using the default settings but got an I/O error 103.
If you have large numbers in your data set this may occur. The reason is that, as part of its normal operation, TWINSPAN adds some very large numbers to each data value, in order to increase the discrimination between them. If you already have large numbers the resulting values may be too large for the data types used within the program. The solution is to transform your data. Usually a square root transformation will resolve the matter.
7. Interpreting TWINSPAN output.
In TWINSPAN OUT: concerning the list of Indicators and their sign (i.e. Species 1 (+), Species 2 (-)), does the (+) represent species present in the class and (-) represent species NOT present in the class? If so, then are the only sample points (quadrats) representing the TWINSPAN classes those listed under the positive group?
In the TWINSPAN output Species1 (+) indicates that species1 is characteristic of the quadrats classified to the right of the centroid of the primary axis of the ordination. It does not mean that the Species1 is absent in all the quadrats to the left of the centroid. An output of, say, Species2 (-) would mean that Species2 is characteristic of quadrats to the left of the centroid. Remember that if TWINSPAN is not undertaken on presence/absence data; it uses pseudospecies so the same species can be an indicator at different levels of abundance.
In the case of presence/absence data there will certainly be a strong tendency for a positive species only to be present in the group of quadrats to the right. However, this is not necessarily always the case.
8. Invalid floating point operation.
This often indicates that there are rows or columns with no data in them. Use the Handling Zeros button on the working data tab to find and remove the offending row or column.
9. My files do not appear to be saved.
This is often caused by the lack of the correct file extension. By default, the Open File dialog shows .csv files. To see other files in the Open File dialog, select .xls, .txt or All Files from the Files of type: drop-down menu. WARNING: the latter option will show all the files in the directory - whether they are compatible with the program or not. If you have saved a file without an extension either add the extension outside the program (open the folder in Windows Explorer or My Computer, single-click the file and press F2 to edit the file name), or open the file and save with the correct extension using 'Save as...'
10. When I press F1, no Help screen appears.
Ensure that the window on which you are seeking help is the active one.
11. I want to use a similarity or distance measure that is not offered by CAP.
Contact PISCES and we will try to implement the measure for you.