I have started coding my first Random Forest Classifier.
My estimators are stored in a 3-dimensional matrix: 120 (number of patients) x 111 (number of sensors) x 5 (values measured by each sensor).
I was thinking about "unrolling" my 3-dimensional matrix into a 2-dimensional matrix to have an array of lists/array of arrays/list of lists per row, something that would look like this:
Sensor_1 ... Sensor_j
Patient_1: [α_11, β_11, γ_11, δ_11, θ_11] ... [α_1j, β_1j, γ_1j, δ_1j, θ_1j]
... ... ... ...
Patient_k: [α_k1, β_k1, γ_k1, δ_k1, θ_k1] ... [α_kj, β_kj, γ_kj, δ_kj, θ_kj]
In this example, α, β, γ, δ, θ correspond to the five values measured by each sensor, j correponds to the sensor's number, and k corresponds to the patient's number.
I am not sure this is the best way to tackle this problem and I would like to know if there us another way to use RandomForestClassifier with a 3-dimensional array.