Removing sample points

Col

Hello,

I have been doing cell spaced laser ablation and I have generated values for a bunch of spots across my sample. I noticed after the DRS step that some values are negative ( as expected). I would like to omit these negative values from the data by either removing them or setting them to a DL so I can use the imaging tool. Using the Iolite imaging tool visually is fine as I can set the scale to ignore those points but when using the region of interest tool to calculate a bulk geochemical value for either the whole grid or a specific mineral, I don't want the values to be calculated with any negative values. Just wondering if I'm missing some easy tool within the software and any help would be appreciated.

Many thanks,
Collette

Joe

Col

While it is possible to replace the data below 0, you may want to reconsider. If you cut off the tail end (i.e. those < 0) of a distribution of measurements, you are artificially bumping up the distribution's mean. If you're sure you want to replace those values, you could do something like

for c in data.timeSeriesList(data.Output):
    c.data()[c.data() < 0] = 0

data.updateResults()
data.emitDataChanged()
data.activeSelectionGroup().changed.emit()

This will loop through all of the output channels and replace data less than 0 with 0.

See also:
This note and this forum post.

All the best,

Col

Hi Joe,

Thank you for your reply. I was wondering if instead of replacing the data below 0 to 0 if I could replace it with the calculated LOD values. I want to avoid artificially bump up the distribution mean but some of the minus values are quite extreme and I think replacing it with the LOD value would be more appropriate. I noted you can export files and replace the data below LOD for each element with 'below LOD' so it has already been calculated for each element, I am just not sure how to assign the value rather than the word 'below LOD' for each element. I was also wondering if this is something I can do within Iolite and not just in an export file? I would like to use those assigned LOD values within the imaging software to generate bulk values using the region of interest tool.

Many thanks in advance,

Collette

Joe

Col

The LOD we normally calculate is for a selection represented by (usually) many points, and the number of points factors into that calculation. The LOD of an individual time-slice is not something we calculate so you'd need to work that part out separately if you want to replace < 0 with that.

Another approach might be to set the values < 0 to be "not a number". In doing so they will not be included in the stats. I think you should be able to do that as follows:

import numpy as np
for c in data.timeSeriesList(data.Output):
    c.data()[c.data() < 0] = np.nan

data.updateResults()
data.emitDataChanged()
data.activeSelectionGroup().changed.emit()

All the best,

Col

Hi Joe,

Thank you, that has really helped me play around with the python workspace.

I noticed that the code above targets all data in the output columns (data.timeSeriesList(data.Output))

If I was just interested in targeting the results in one column within the output e.g. Pb207/Pb206 how would I go about it?

I attempted:

data.timeSeriesList(data.'Pb207/Pb206')
data.timeSeries('Pb207/Pb206')

(Background information - I noticed that the code above works really well for the trace element issue I was having but I noticed that the Pb207/Pb206 DRS was calculating ratios from values that should have been removed from the beginning. I altered the DRS you gave me and added an a mask so the ratios would never be generated but interested in seeing how removing the values in the column before 0 and above 1 works as well. I noticed by targeting the entire output list it altered the Pb207/Pb206 ratios which I would like to avoid).

Many thanks,
Collette

Joe

Col

By Pb207/Pb206 DRS do you mean the U-Pb one or is this a Pb isotope DRS?

You can get an individual channel with data.timeSeries('ChannelName'), but keep in mind that any time you run the DRS, that channel will be removed and recreated from the input data. Do you really want to mask the intermediate channel 'Pb207/Pb206' or do you want to do the final/output one? You could replace the loop in the original code snippet with something like:

import numpy as np
c = data.timeSeries('Pb207/Pb206')
c.data()[c.data() < 0] = np.nan

data.updateResults()
data.emitDataChanged()
data.activeSelectionGroup().changed.emit()

If you haven't seen it already, this thread may be of interest.

All the best,