Should be very fun!
|
|
|
Saturday, February 19, 2022
Range-Based Volatility Seminar, MONDAY 2/21
Saturday, February 12, 2022
Deep Recurrent Neural Nets with Long Short Term Memory
Again. LTSM may be emerging as very big deal in recurrent NN modeling. I blogged on it before (e.g. here) but I still don't understand it deeply. Does anyone?
Maybe it's just a device for avoiding the vanishing-gradient problem (not that that isn't important); maybe it's more.
This new paper is very well done and features LSTM prominently.
| By: | Lars Lien Ankile; Kjartan Krange |
| Abstract: | This paper presents an ensemble forecasting method that shows strong results on the M4Competition dataset by decreasing feature and model selection assumptions, termed DONUT(DO Not UTilize human assumptions). Our assumption reductions, consisting mainly of auto-generated features and a more diverse model pool for the ensemble, significantly outperforms the statistical-feature-based ensemble method FFORMA by Montero-Manso et al. (2020). Furthermore, we investigate feature extraction with a Long short-term memory Network(LSTM) Autoencoder and find that such features contain crucial information not captured by traditional statistical feature approaches. The ensemble weighting model uses both LSTM features and statistical features to combine the models accurately. Analysis of feature importance and interaction show a slight superiority for LSTM features over the statistical ones alone. Clustering analysis shows that different essential LSTM features are different from most statistical features and each other. We also find that increasing the solution space of the weighting model by augmenting the ensemble with new models is something the weighting model learns to use, explaining part of the accuracy gains. Lastly, we present a formal ex-post-facto analysis of optimal combination and selection for ensembles, quantifying differences through linear optimization on the M4 dataset. We also include a short proof that model combination is superior to model selection, a posteriori. |
| Date: | 2022–01 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:2201.00426&r=&r=for |
Wednesday, February 9, 2022
Climate Policy Postdoc
The the newly established FutureLab at the Potsdam Institute for Climate Impact Research (PIK) is looking to hire a postdoctoral researcher on causal climate policy analysis.
The successful candidate will be involved in a joint research project with Nicolas Koch (MCC and IZA) and Felix Pretis (University of Victoria and University of Oxford) that seeks to provide a global and cross-sectoral causal evaluation of effective climate policies. We expect the candidate to co-lead policy-oriented and data-intensive econometric research that builds on and further develops our prior work. The goal is to pair machine learning with program evaluation tools to estimate causal treatment effects in settings in which standard methods are limited.
Criteria:
- PhD (or be close to finishing their PhD) in economics or a related field,
- strong skills in econometric methods as well as expertise and research interest in environmental and climate change economics,
- strong coding skills are a prerequisite, especially in R
More information is available here: https://www.pik-potsdam.de/de/aktuelles/stellen/postdoctoral-position-on-causal-policy-analysis201d-m-f-d
Application deadline: March 15th, 2022
Feel free to reach out to Nicolas Koch (https://www.mcc-berlin.net/en/about/team/koch-nicolas.html) or myself for more details on the position.