Predicting future transporation mode based on previous patterns I want to predict future transportation mode (car,walk, bus) of a user in a trip. A trip is composed of multiple segments and each segment consists of features like speed, distance, time and transportation mode. Based on known features of previous segments (here segment 1,2,3) i want to predict the transportation mode or if possible predicting the features of next segment (segment 4) and so on for segment 5,6 etc. 
segment No -> speed -> distance -> time -> transportation mode
segment 1 -> 70 km/h -> 30 km -> 28 min -> car
segtment 2 -> 3 km/h -> 1 km -> 15 min -> walk
segment 3 -> 40 km/h -> 10 km -> 20 min -> bus
segment 4 -> ??? -> ??? -> ??? -> ???
segment 5 -> ??? -> ??? -> ??? -> ???
.
.
.
.
.
segment n -> ??? -> ??? -> ??? -> ???
which statistical or machine learning models would be helpful to predict the future transportation modes based on existing patterns?
 A: I assume you have a dataset of Trips, which consists of $N$ trips, each one composed of $n_i$ segments data-pairs ($\boldsymbol{x^{(i,j)}}$, $y^{(i,j)}$), where $\boldsymbol{x}$ is your vector of features (distance, speed, time) and $y$ is your transportation mode. 
Only based on the structure of your data, the obvious possibility that comes to my mind is a simple MLP to map your trip features to transportation mode ($\boldsymbol{x}^{(i,j)} \to y^{(i,j)}$) -- that's the classification part of your problem -- and you could try to use a Reccurent Neural Net (e.g. LSTM) for the prediction problem of inferring a sequence of future trip segments based on previous ones $[\boldsymbol{x}^{(i,1)}, \boldsymbol{x}^{(i,2)}, \boldsymbol{x}^{(i,3)}] \to [\boldsymbol{x}^{(i,4)}, \cdots, \boldsymbol{x}^{(i,n)}]$.
Of course there are a lot of methods and tricks out there, and choosing which ones fit best your needs is a difficult question that also depends on your goals, your dataset size, your computing ressources, etc.
