I would like to derive a pattern that tells me when the door should be open and when closed. For instance, if the status spectrum refers to the front door and recorded data show that the first day it is opened for 1 minute at 9am, at 12 noon and at 6 pm, and that the second day it is opened for 1.5 mins at 9.30, 12.30, and 6.30, and the third day… similarly, then there should be derive a pattern where
the front door is opened for less than, say, two minutes every day between 9 and 10, between 12 and 1 pm, and between 6 and 7 pm (or something similar).
How to do it? Any algorithms? Can this be done using weka or other machine learning programs?
EDIT:
You have to define it better for any machine learning algorithm to be applicable.
For supervised learning, the algorithm tries to predict a “correct” label.
What is a correct label here? What is the cost of misprediction?
Alternatively, for unsupervised leaning, the algorithm tries to create useful subsets of the data. What is a good subset here?
Are there any independent variables?