I want to train a neural network to predict what my favourite home-work route will be for a particular day. I have these features for routes on a day:
temperature, humidity, congestion, distance, duration
I have come up with this concept of training/testing a network:
// // training features result in route 1,2 or 3 // Network.train([30,10,12,20,12] , 1) Network.train([20,10,22,20,14] , 3) Network.train([23,10,2,20,10] , 2) Network.train([20,10,22,20,12] , 2) // // On a new day, predict which route the user is most likely to take: // var route = Network.test([25,8,12,22,12])
My question is: is this a viable approach? Can I make relevant predictions this way if I have enough training data?
Can I generate an outcome between 1 and 3 this way?