Classification Learner Input Data format

How does multidimensional input data (multiple long vectors) need to be formatted so Classification Learner can train on it, and how do you attach categories for classification to this data?

Scenario: I have multiple logs of power draw from various appliances in my house. I have data on my toaster, blender, fridge, lights, etc. For each appliance, I have ~50 data vectors, each representing one day of usage. Each day of use is 15-20k data points long, but not every day's worth of data is the same length (a quirk in my logging).

Ideally I'd assemble a structure with fields "toaster", "blender", etc, and each field would be a cell array of length 50 (days), where each cell contained one vector of length 15-20k, or each field could be an array of 50 columns, where each column was one (variable length) day's worth of data. But there isn't great documentation on how this kind of data is ingested into the CL App.

Basically, how do I package each appliance's 50 days of data into a variable, associate the identifier (appliance name) with it's chunk of 50 days, and then run this through the classification learner so it can predict which appliance a data set (one day's worth of data) represents?

Tables have "one restriction that each variable must have the same number of rows" so I don't know how that would work. Other

[madhan ravi]

Upload your code and the files required.

[Tim Jenks]

See attached for an example data file. I'm hoping to import into classification learner and train a ML algorithm such that when I give it a 1.5k-1.7k length vector, it could tell me if it's a toaster, fridge, water heater, or lights. (I cut the vectors down by an order of magnitude for file size.) I've scoured the documentation but it seems sparse on how to properly format data like this for ingestion. Thanks.

[madhan ravi]

describe which data you want to extract?

[Tim Jenks]

I don't understand your question. The alldata variable contains 4 cell arrays, each array is one appliance. Each appliance contains 50 data sets (double vectors) characteristic of its power draw during a day. I want to train an algorithm that will recognize any double vector as one of these appliances. a.k.a after training the ML algorithm I give it a 16,000-length vector, and it outputs "toaster", "fridge", "heater", or "lights".