[PDF] Machine Learning: An Applied Econometric Approach

2017 Jeep Wrangler Owners Manual - Microsoft

higher center of gravity, if this vehicle is out of control it may roll over while some other vehicles may not Do not attempt sharp turns, abrupt maneuvers, or other

2017 WRANGLER - Jeep

with determined perseverance , far-reaching capability enjoy best-in-class off- road performance 6 with the 2017 jeep ® wrangler unlimited rubicon ® in firecracker red

2017 Jeep Compass Owners Manual - Microsoft

higher center of gravity, if this vehicle is out of control it may roll over while some other vehicles may not Do not attempt sharp turns, abrupt maneuvers, or other

2017 Jeep Patriot Owners Manual - Dealer eProcess

2017 VEHICLES SOLD IN CANADA With respect to any Vehicles Sold in Canada, the name FCA US LLC shall be deemed to be deleted and the name FCA

Machine Learning: An Applied Econometric Approach

Journal of Economic Perspectives—Volume 31, Number 2—Spring 2017—Pages 87–106 M achines are increasingly doing “intelligent” things: Facebook recognizes faces in photos, Siri understands voices, and Google translates websites The fundamental insight behind these breakthroughs is as much statis-tical as computational

2017 Jeep Grand Cherokee Owners Manual

this Owner’s Manual will help assure safe and enjoyable operation of your vehicle After you have read the Owner’s Manual, it should be stored in the vehicle for convenient

2017 Jeep Renegade Owners Manual - Dealercom US

higher center of gravity, if this vehicle is out of control it may roll over while some other vehicles may not Do not attempt sharp turns, abrupt maneuvers, or other

2017 PATRIOT - Jeep

2017 patriot every season renews the adventure enter the world of the 2017 jeep

[PDF] agrément jeunesse éducation populaire

[PDF] calcul sensibilité capteur

[PDF] journée du patrimoine 2017 paris

[PDF] sensibilité spécificité medecine

[PDF] retrouver numéro agrément jeunesse et sport

[PDF] vpp vpn

[PDF] journée du patrimoine paris

[PDF] fonjep

[PDF] préparation agrégation biochimie génie biologique

[PDF] journee du patrimoine

[PDF] trouble du spectre de l'autisme symptome

[PDF] autisme cours pdf

[PDF] exemple fiche machine

[PDF] l'autisme définition pdf

[PDF] exemple dossier machine maintenance

Journal of Economic Perspectives - Volume 31, Number 2 - Spring 2017 - Pages 87-106 M achines are increasingly doing “intelligent" things: Facebook recognizes faces in photos, Siri understands voices, and Google translates websites . The fundamental insight behind these breakthroughs is as much statis- tical as computational. Machine intelligence became possible once resear chers stopped approaching intelligence tasks procedurally and began tackling t hem empirically. Face recognition algorithms, for example, do not consist of hard-wired rules to scan for certain pixel combinations, based on human understandi ng of what constitutes a face. Instead, these algorithms use a large dataset o f photos labeled as having a face or not to estimate a function f (x) that predicts the pres-

ence y of a face from pixels x. This similarity to econometrics raises questions: Are these algorithms merely applying standard techniques to novel and large

datasets? If there are fundamentally new empirical tools, how do they fit with wha t we know? As empirical economists, how can we use them? 1 We present a way of thinking about machine learning that gives it its own place in the econometric toolbox. Central to our understanding is that machine learning 1 In this journal, Varian (2014) provides an excellent introduction to many of the more no vel tools

and “tricks" from machine learning, such as decision trees or cross-validation. Einav and Levin (2014)

describe big data and economics more broadly. Belloni, Chernozhukov, and Hanson (2014) present an econometrically thorough introduction on how LASSO (and close cousins) can be used for inference in

high-dimensional data. Athey (2015) provides a brief overview of how machine learning relates to causal

inference.

Machine Learning: An Applied

Econometric Approach

Sendhil Mullainathan is the Robert C. Waggoner Professor of Economics and Jann Spiess is a PhD candidate in Economics, both at Harvard University, Cambridge, Massachusetts.

Their email addresses are mullain@fas.harvard.edu and jspiess@fas.harvard.edu. For supplementary materials such as appendices, datasets, and author disclosure statemen ts, see the article page at https://doi.org/10.1257/jep.31.2.87 doi=10.1257/jep.31.2.87

Sendhil Mullainathan and Jann Spiess

88 Journal of Economic Perspectives

not only provides new tools, it solves a different problem. Machine lear ning (or rather "supervised" machine learning, the focus of this article) revolves around the problem of prediction: produce predictions of y from x. The appeal of machine learning is that it manages to uncover generalizable patterns. In fact, the success of machine learning at intelligence tasks is largely due to its ability to discover complex structure that was not specified in advance. It manages to fit complex a nd very flex ible functional forms to the data without simply overfitting; it finds functions that work well out-of-sample. Many economic applications, instead, revolve around parameter estimation: produce good estimates of parameters that underlie the relationship between y and x. It is important to recognize that machine learning algorithms are not built for this purpose. For example, even when these algorithms produce regres sion coef ficients, the estimates are rarely consistent. The danger in using these tools is taking an algorithm built for y ˆ , and presuming their ˆ have the properties we typically associate with estimation output. Of course, prediction has a long histo ry in econo metric research - machine learning provides new tools to solve this old problem. 2 Put succinctly, machine learning belongs in the part of the toolbox marked y ˆ rather than in the more familiar ˆ compartment. This perspective suggests that applying machine learning to economics re quires finding relevant y ˆ tasks. One category of such applications appears when using new kinds of data for traditional questions; for example, in measuring e conomic activity using satellite images or in classifying industries using corpo rate 10-K filings. Making sense of complex data such as images and text often involves a prquotesdbs_dbs7.pdfusesText_5