The instructions say that selecting "Data->Remove Bad Data Points" from the menu will open the data reader and allow me to select points, that can then be deleted by pressing "enter". The data reader opens, I can select points, but pressing "enter" while a point is selected does nothing. Neither does pressing "return", "delete", "backspace" or "space".

How do I delete bad data points without having to go to the (several megabyte) worksheet and try and find individual points by hand?

An additional option was brought to my attention:

If the Remove Bad Data menu command and the Masking tool are too cumbersome, you could try using Analysis:Extract Worksheet Data. For instance, suppose you have data in a column called TEST that ranges from a value of 1 to 105. For some reason you have determined the data between 80 and 105 to be outliers (erroneous data) and you would like to remove these outliers from your graph. Prior to plotting, perform the following in the worksheet:

1. Select Analysis:Extract Worksheet Data.
2. Enter col(TEST)[ii]<80 in the space provided.
3. Click OK.

Any rows of data that satisfy the criteria you have entered will be extracted and placed into a new worksheet from which you can plot "outlier free". I would like to reiterate the fact that I said rows since there may be columns in the worksheet which you do not want affected. If this is the case, you should separate the columns you do want to affect from the columns you don't want to affect by moving either group into a new worksheet.

Just thought I'd let you know!

Sincerely,
Ryan Toomey

couldn´t notice any problems (Origin6.0-SR1), to delete bad data-points, using the Return-key. Yet, I also noticed, that it seems possible, to delete such a point simply by double-clicking on it, after selecting that menu and getting that data-reader- cursor. Maybe, you can get rid of single points with that workaround?

I've played with masking- including trying to work through the document you included a pointer to- but no matter what I do the buttons in the mask toolbar remain ghosted.

Increasing the bisection search points works on the smaller plots, although it seems I can only remove one point at a time- you select the menu entry, the reader appears, you either doubleclick or press enter on a point and it disappears. Along with the reader. I have many points to remove- is there a way of keeping the reader open?

For the larger plots, although I can now delete points, I can't select the correct points- often the program selects a point 50% across the graph from where I click. I've discovered that the "bisection search points" control is limited to areound 40K points. My large datasets that I want to work with are 100K points. Is there a way to override the limit?

grahamc,

Your questions/comments are indented, bolded and in italics. My answers/thoughts follow each question/comment:

I've played with masking- including trying to work through the document you included a pointer to- but no matter what I do the buttons in the mask toolbar remain ghosted.

The buttons on the Mask toolbar are restricted to use with Line + Symbol or Scatter plots only. This limitation is indicated in the beginning of the masking article. Specifically, it states the following: "When affecting a graph window, the new masking features can be applied to active Scatter and Line + Symbol plots, but are technically not supported for any other plot types." If you require a line plot for your work, change the plot to a Line + Symbol plot and the Symbol Size to 0. Masking can still be performed on this plot type even though the Symbol Size is 0.

Increasing the bisection search points works on the smaller plots, although it seems I can only remove one point at a time- you select the menu entry, the reader appears, you either double-click or press enter on a point and it disappears. Along with the reader. I have many points to remove- is there a way of keeping the reader open?

The Remove Bad Data operation is only designed to handle one point at a time. A suggestion to automate this procedure for multiple data points has been entered in our database. Thanks for the idea! In the meantime the only way to repeat this procedure without having to select it again would be to write a script.

For the larger plots, although I can now delete points, I can't select the correct points- often the program selects a point 50% across the graph from where I click. I've discovered that the "bisection search points" control is limited to around 40K points. My large datasets that I want to work with are 100K points. Is there a way to override the limit?

The Bisection Search Points control has a limit of (2^16)-1 or 65535 points in the release version of Origin 6.0 and Origin 6.0 SR1. This limitation was noted and increased to 999,999 for Service Release Patch 2 (due out sometime in March or April of 2000). Your name has been added to the report so that we can contact you when it becomes available (assuming your e-mail address remains the same). Until then, the only reliable way to zero in on a "bad" data point would be to click as closely as you can to it and use the arrow keys on your keyboard to do the rest of the work. Unfortunately this method is very slow!

I hope I have answered all of your questions.

Sincerely,
Ryan Toomey

The Data:Remove Bad Data Points menu command should indeed allow you to remove a bad data point upon selecting the data point and pressing the ENTER key. In fact, in all my tests using this feature it did just that. Since I am unable to reproduce the problem, I am led to believe that the operation may be failing as a result of the large size of your dataset. To test this theory, try increasing the Bisection Search Points value in Tools:Options => Miscellaneous tab to a number higher than the number of points you have in your data plot. Once you have done this, try removing a bad data point again. Alternatively, see my answer to your suggestion for Origin 7.0 below.

Although it is not possible to "drag-select a region on the graph and delete all the data points in that region" in Origin 6.0 (SR1 included), it is possible to mask a data point or region of data in the active dataset. Once a data point or region of data is masked, it is ignored during all applicable analysis routines and can even be hidden from view when plotting. Please take a look at the July 1999 Technical Review article for more information on these and other masking features.

Sincerely,
Ryan Toomey