This forum is about

Data mining with eye tracking

Posted by:
**
Sasa
**

Date: November 18, 2019 09:04AM

Hello,

I will be conducting an eye tracking study with two different groups of students learning something on a computer screen. I want to measure how the two groups differ in the sequences of fixations on specific areas of interest. Let's assume I have 5 areas of interest and 10 participants per group.

What kind of algorithm do I need to find out if the participants within the groups show specific fixation patterns?

Thanks!

I will be conducting an eye tracking study with two different groups of students learning something on a computer screen. I want to measure how the two groups differ in the sequences of fixations on specific areas of interest. Let's assume I have 5 areas of interest and 10 participants per group.

What kind of algorithm do I need to find out if the participants within the groups show specific fixation patterns?

Thanks!

Posted by:
**
webmasterphilfv
**

Date: November 26, 2019 07:16AM

Hi,

I think there would be many ways of doing such experiments.

- First, if it is for some psychological studies, I know that psychologists like to use statistical test to compare two groups. For this approach, it would require to find some appropriate test to measure the significance of differences between the two groups.

- If you want to use a more data mining approach, you could see the sequences of fixations as sequences of symbols and then apply a software like SPMF and some algorithm for sequential pattern mining to find the frequent sequence of fixation areas. Then, you could do this for each group and then compare the patterns that you have found in both group to see if they are different. Then, you could again use some statistical test, or some other measures that are more simple, or just compare by hand to see if the patterns are different.

- There are some techniques in sequential pattern mining that specifically aim at comparing the patterns of two grous. Those are called emerging pattern mining or contrast pattern mining, but you would have to use this for sequential pattern mining since your dataset is sequences...

That is just my opinion after thinking about it for a few minutes. There are certainly other possibilities.

It is an interesting topic.

Best regards,

I think there would be many ways of doing such experiments.

- First, if it is for some psychological studies, I know that psychologists like to use statistical test to compare two groups. For this approach, it would require to find some appropriate test to measure the significance of differences between the two groups.

- If you want to use a more data mining approach, you could see the sequences of fixations as sequences of symbols and then apply a software like SPMF and some algorithm for sequential pattern mining to find the frequent sequence of fixation areas. Then, you could do this for each group and then compare the patterns that you have found in both group to see if they are different. Then, you could again use some statistical test, or some other measures that are more simple, or just compare by hand to see if the patterns are different.

- There are some techniques in sequential pattern mining that specifically aim at comparing the patterns of two grous. Those are called emerging pattern mining or contrast pattern mining, but you would have to use this for sequential pattern mining since your dataset is sequences...

That is just my opinion after thinking about it for a few minutes. There are certainly other possibilities.

It is an interesting topic.

Best regards,