This week, I am working to get the dynamic time warping functionality into my program. The process of doing so includes re-processing the features to include the time series, putting each series back together when we construct sequences, and then performing the DTW to generate a number that will be used to compute the kNN of each sequence which can then be used for predictions with the models. The processing time of these activities has gone up significantly since we have been using five different metrics with each of the F phase datasets. I am returning to school next week, and once I've completed the DTW processing all that will remain before we put together our second paper (The date for the reach journal we would like to submit it to is October 1), I am hoping I will have time to look again into the Agglomerative Hierarchical Clustering concept, which I did not successfully complete when we explored it earlier in the summer and then changed focus to the paper. We heard back
I have been looking at the way a subject's features change over the duration of their participation in the study. To do so, I am exploring several different routes of measuring change. The first is to compare daily metrics to the average of those of the first three 'baseline' days. These are the days that the subject wore the actigraph watch but did not receive light treatment. The second is to compare one day's metrics to the previous day's, and the final is to compare a day to the average metrics of all the days preceding it. We are uncertain as to which of these may be most helpful going forward.
In order to avoid redundancy, I am going to only look at a few selected features for each subject: sleep, wake and transitions for lights on and lights out, longest bout, sleep onset latency and daytime activity ratio.
I also want to represent each of the changes in several different ways: the raw difference between the day's value and the original value, the percent change and the sign of the change (-1 if the day's value is lower than the original value, 1 if it is lower and 0 if there was no change). The most useful of these going forward will likely be the third, as this is the way that we would like to ultimately represent values within our majority-voting model.
Comments
Post a Comment