For the past year, I have spent the majority of my free time learning a variety of ML techniques (boosting, random forests, neural nets, SVMs etc.), but I have not been able to find a lot of material (books, papers etc.) that explain ML techniques in the presence of time-dependent data. Are there any textbooks or other resources on this?
$\begingroup$
$\endgroup$
4
-
$\begingroup$ Time-dependent data can be modeled with a wide range of models. Could you provide us more information on what kind of models are you interested? You could also have a look here: cs.stackexchange.com/questions/13937/… $\endgroup$– Yannis AssaelCommented Jun 11, 2015 at 2:30
-
$\begingroup$ I am interested in doing this for random forests and boosted decision trees. I found some information pertaining to neural nets and time-dependent data, but I had trouble finding information for boosting and RFs. $\endgroup$– mmmmmmmmmmCommented Jun 11, 2015 at 2:34
-
$\begingroup$ Thank you for the link. I will certainly check out those resources! In your experience, do time series models (ARIMA, GARCH etc.) tend to outperform ML techniques?? $\endgroup$– mmmmmmmmmmCommented Jun 11, 2015 at 2:35
-
1$\begingroup$ From my experience, and given that you have a marked amount of data, I would definitely go for Recurrent Neural Networks, starting with simple RNNs and moving to LSTMs, SCRNNs, or GRUs. They are very powerful in modeling time-series for either classification or regression tasks. P.S. there is also an ARIMA-RNN, while ARIMA, GARCH are considered ML as well. $\endgroup$– Yannis AssaelCommented Jun 11, 2015 at 2:41
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
This seems like a good place to start.
"Machine Learning Strategies for Time Series Forecasting" by Gianluca BontempiSouhaib Ben TaiebYann-Aël Le Borgne
The increasing availability of large amounts of historical data and the need of performing accurate forecasting of future behavior in several scientific and applied domains demands the definition of robust and efficient techniques able to infer from observations the stochastic dependency between past and future. The forecasting domain has been influenced, from the 1960s on, by linear statistical methods such as ARIMA models. More recently, machine learning models have drawn attention and have established themselves as serious contenders to classical statistical models in the forecasting community. This chapter presents an overview of machine learning techniques in time series forecasting by focusing on three aspects: the formalization of one-step forecasting problems as supervised learning tasks, the discussion of local learning techniques as an effective tool for dealing with temporal data and the role of the forecasting strategy when we move from one-step to multiple-step forecasting.
