Latest update

- Clarification on explanation of mechanics of tensor product
- Termwise max (resp. min) of two convergent (resp. divergent) series
- What happens to Clifford algebra structure and periodicity when the field is weird?
- Stalk of the sheafification
- Dihedral groups textbook example
- Inequalities in one variable: Total revenue
- Upper bound on Frobenius Coin Problem
- Which of the following rules are operations on the indicated set? $a*b$ = $a.b$ and $a*b$=|$a^2$-$b$|?
- Functions & Relations
- Neat result about the average integral of a converging function
- How to graph a curve finding the Cartesian equation
- Sign of function with c.d.f. and p.d.f. of normal distribution
- Tensor equation with cross product
- Propogation of two sets of errors in a matrix formula
- Understanding Polig-Hellman algorithm for DLP via example
- Application of Fatou's Lemma
- Problem on conditional expected value with to a random variable
- find equation in other basis
- Definition of a `flatten' map
- If $F$ and $G$ are biadjoint, how is $\operatorname{Nat}(F,F)\simeq\operatorname{Nat}(G\circ F,1)$?

# Multi-Source Time Series Data Prediction

2017-07-17 11:41:05

I was wondering if anyone has experience with temeseries prediction for data from. multiple sources. So for instance, timeseries a,b,..,z each have their own shape, some may be correlated with others. the ultimate goal is to have a model trained such that the t+1 value for any given data source can be predictied.

I personally have two solutions that in theory could work, but was wondering if anyone knew of other frequently used methods.

1) Multi-task learning with LSTM 2) use feature engineering to model properties of each timeseries source as features along with usual features and use these with LSTM