6 ms·
Lectures in Quantitative Economics as Python and Julia Notebooks
- westurner 8y agoThere are undergraduate and graduate courses in each language: Python version: https://lectures.quantecon.org/py/ https://lectures.quantecon.org/py/ Julia version: https://lectures.quantecon.org/jl/ https://lectures.quantecon.org/jl/
- bigmit37 8y agoJust gave it a quick look— Loaded with lots content. Thank you so much for sharing.
- kaffee 8y agoDoes anyone else find it strange that there is no real-world data in these notebooks? It's all simulations or abstract problems. This gives me the sense, personally, that economists aren't interested in making accurate predictions about the world. Other fields would, I think, test their theories against observations.
- wjnc 8y agoIt's an educational course in quantitative economical methods. Fitting real world data is messy and would probably distract. There obviously is overlap with metrics but as an undergrad course I'd separate this too. They do have ample links to scientific papers that do use real world data. There's a pages long list of references [1]. Do check them out if you're into economical science. [1] https://lectures.quantecon.org/jl/zreferences.html https://lectures.quantecon.org/jl/zreferences.html
- kome 8y agoyeah, doing real empirical stuff is messy. but that's exactly why people should learn on real data, and not in a sandbox.
- wjnc 8y agoI let my kids practice with hammers, nails and wood (tools / supplies) before I introduce building a piece of furniture (the educational end-goal). These models are the tools of the trade. I agree with the sentiment that if work is messy, teaching should have messy as well. But not when you're starting out with new tools.
- rezahandzalah 8y ago+1 for the kids tools. To extend, I let my kids practice hammering first, then sawing, then some other skill. Learning to work with messy data can wait until you are used to the new tools.
- zwaps 8y agoIt is literally impossible to run these models on data without first understanding these methods deeply, because causal inference for these observational data IS extremely difficult. If you do not understand your model deeply, or if you do not understand your data deeply, you are likely producing garbage. This course relates to the first point. There's a lot of structural econometric papers that do exactly what you ask, but you need graduate level statistics and a deep understanding of discrete choice, identification and simulation methods. Structural econometrics is a field where PhD students, in their 5 year of study, usually produce only one complete study, if that.
- yyyymmddhhmmss 8y agoYou’re pretty much right on the nose. Of course, it depends on who they work for. Effectively, the American field of economics is an exercise in decoupling private reality from public theory.
- python_gt_r 8y agoAnd wait until they fit 12 parameters on 20 observations where each of the observations has as well a massive measurement error.
- westurner 8y agopandas-datareader can pull data from e.g. FRED, Eurostat, Quandl, World Bank: https://pandas-datareader.readthedocs.io/en/latest/remote_data.html https://pandas-datareader.readthedocs.io/en/latest/remote_da... pandaSDMX can pull SDMX data from e.g. ECB, Eurostat, ILO, IMF, OECD, UNSD, UNESCO, World Bank; with requests-cache for caching data requests: https://pandasdmx.readthedocs.io/en/latest/#supported-data-providers https://pandasdmx.readthedocs.io/en/latest/#supported-data-p... The scikit-learn estimator interface includes a .score() method. "3.3. Model evaluation: quantifying the quality of predictions" https://scikit-learn.org/stable/modules/model_evaluation.html https://scikit-learn.org/stable/modules/model_evaluation.htm... statsmodels also has various functions for statistically testing models: https://www.statsmodels.org/stable/ https://www.statsmodels.org/stable/ "latex2sympy parses LaTeX math expressions and converts it into the equivalent SymPy form" and is now merged into SymPy master and callable with sympy.parsing.latex.parse_latex(). It requires antlr-python-runtime to be installed. https://github.com/augustt198/latex2sympy https://github.com/augustt198/latex2sympy https://github.com/sympy/sympy/pull/13706 https://github.com/sympy/sympy/pull/13706 IDK what Julia has for economic data retrieval and model scoring / cost functions?
- ploika 8y agoThat's an ironic conclusion to draw. You could test your own theory against observations that calculations with real world data are very much a part of economics, but are just not part of this particular course.
- theoh 8y agoThere are serious critiques of economic theory out there, which tend to say that kind of thing. But if you compared these notes to the notes for a college level physics course, you would find a similar level of abstraction, idealized models, and absence of real world data. Those things are not in themselves indicators that physicists (or economists) don't care about the real world. In any mature field, there is a body of knowledge and techniques to be learnt. There's a certain formalism to be picked up, rather than just staring at data. There might be legitimate reasons for dismissing the general approach taken by mainstream economic theory, but what you seem to be saying ("hmmm, my intuition is that this stuff doesn't focus enough on accurately predicting the real world") is not a reasoned critique.
- cirgue 8y ago> This gives me the sense, personally, that economists aren't interested in making accurate predictions about the world. Other fields would, I think, test their theories against observations. You say this as though using mock-up data to teach techniques isn't a universal practice in literally every other discipline.
- Mikeb85 8y agoIt's an academic course. You learn using models and basic concepts, then eventually apply it to real data. Ever taken statistics courses? You're not doing multiple regression analysis on real world data on day 1. On day 1 you're learning odds using playing cards and coin flips.
- BeetleB 8y ago>Ever taken statistics courses? You're not doing multiple regression analysis on real world data on day 1. On day 1 you're learning odds using playing cards and coin flips. Curiously enough, my undergrad statistics textbook was loaded with problems where the data was taken straight from a journal paper. The book has poor reviews on Amazon, but I think it's the best I've seen. https://www.amazon.com/Probability-Statistics-Engineering-Sciences-Devore/dp/0538733527/ref=sr_1_1?ie=UTF8&qid=1549388004&sr=8-1&keywords=Statistics+Jay+Devore https://www.amazon.com/Probability-Statistics-Engineering-Sc...
- BeetleB 8y ago>Does anyone else find it strange that there is no real-world data in these notebooks? It's all simulations or abstract problems. Pretty much every course I took in undergrad physics had no real world data. The intro level courses were especially fun, when we'd go into the lab and get such horrible data that we'd never conclude what they're teaching in the theory classes. We wondered what the point of the lab even was. The biggest offender is the friction model. Heck no - it's not proportional to the normal force. No one could successfully show that in the lab. And a quick Google search shows you a trivial experiment where just changing the orientation and keeping the normal force the same leads to wildly different frictions.
- evrydayhustling 8y agoIt's amazing how we are watching use cases for notebooks and spreadsheets converging. I wonder what the killer feature will be to bring a bigger chunk of the Excel world into a programmatic mindset... Or alternatively, whether we will see notebook UIs embedded in Excel in the future in place of e.g. VBA.
- b_tterc_p 8y agoThat’s not a bad idea. Spreadsheets are pure functional languages built that use literal spaces instead of namespaces. Notebooks are cells of logic. You could conceivably change the idea of notebook cells to be an instance of a function that points to raw data and returns raw data. Perhaps this just Alteryx though
- FabHK 8y agoNotebooks are very much based on linear order and side-effects, though (while Excel is, as you say, a purely functional DAG). Not sure how they can converge.
- jacobolus 8y ago> Notebooks are very much based on linear order and side-effects Try https://beta.observablehq.com/ https://beta.observablehq.com/
- b_tterc_p 8y agoTwo types of cells. One type defines functions. Another are instances of those functions that require inputs. Arrange them in click n drag floaty space. Cells of type one exist in a name space. Cells of type two exist in spacey space. This kind of thing exists at a larger scale for pipeline visualization. I could see it working for notebooks.
- 20190205 8y agoThis is brilliant. I'm picturing the ability to write a Python function with the parameters being just like the parameters in an Excel function. You can drag the cell and have it duplicated throughout a row, updating the parameters to correspond to the rows next to it. It would exponentially expand the power of excel. I wouldn't be limited to horribly unmaintainable little Excel functions. VBA can't be used to do that, can it? As far as I understand (and I haven't investigated VBA too much) VBA works on entire spreadsheets. Essentially, replace the excel formula `=B3-B4` with a Python function `subtract(b3, b4)` where Subtract is defined somewhere more conveniently (in a worksheet wide function definition list?).
- projectramo 8y agoI'm jealous of present day students. I wish I had these tools when I was a student (lectures laid out as notebooks that you can interact with to see how the graph changes). Of course just reading through or listening to clear explanations is still key.
- albertshin 8y agoHi, present day (grad) student here that has been seeing this change happen gradually over my academic career. honestly there have been times I wish that technology stayed out of education because I feel like that the clear explanations have disappeared in exchange for cool graphics or videos (maybe a budget reallocation to the design/graphics team on the publisher's side?). or a related thing is I've noticed in class discussion happens less as slides have replaced the whiteboard/chalkboard as the class speeds through pre written formulas or texts. overall, perhaps there's a case for more quantity of info being relayed thanks to tech but I feel like quality has suffered as a result. of course ymmv according to prof and institution
- sinuhe69 8y agoThe problem is not the tech but the learning style. With interactive notebooks and ton of other materials available, students can participate in pro-active reading before the class so the actual class time can be used more for discussion, not just lecturing. Sadly, most students and professors still do the old way of passive listening.
- 4thaccount 8y agoI can't possibly see how any student in certain subjects like engineering can do much reading ahead. I listened to the lecture, tried to study notes, did insane amounts of homework which always took hours, worked on projects, and then had to study for quizes and tests. I would've loved to check out a chapter before class, but what little time I had was for sleeping, eating, and a little socializing. Granted there are some people that are way smarter than me or especially if you had an easy program you might have been able to read ahead. I like the concept of Notebooks a lot, but you have to be careful that students aren't getting slightly flashier presentations that come with confusing installation woes.
- nwhatt 8y agoMight as well call intro to machine learning, many models overlap.
- zwaps 8y agoProbability theory, optimization, statistics and so forth do not differ between economics and computer science, so it makes sense they are the same. You would see a difference in that these sort of models are used for causal inference and counterfactual analysis, whereas Machine Learning is mostly predictive. That being said, Machine Learning is starting to apply methods developed in econometrics and/or stats, like GMM and Time Series methods. For example, Long-Term Memory models are quite recent additions to Machine Learning. The short-memory process restriction of autoregressive models has been worked on since the early 80's.
- openloop 8y agoPretty cool actually.
- abdullahkhalids 8y agoDoes anyone have suggestions on how to go about monetizing an interactive course like this? And any comments on how successful it could be?
- rmbeard 8y agoLook at the data camp model.
- thoughtstheseus 8y agoDoes anyone have a sense of how useful an economics background is for data science? I've heard mixed things.
- logancg 8y agoI'm econ undergrad -> DS -> Machine learning. Econ is very useful for data science if you focus on the right subjects: statistics, math, and experimental design. You get all the hard skills you need to interact with data that a statistician or computer scientist gets, with the (significant, unique) benefit of learning how to ask the right question or design the right experiment given what is likely a messy, weird, social scientific question. On the other hand, if you don't do any quantitative, empirical, or experimental economics -- i.e. you only do theory or political econ -- then you won't pick up these skills (as much).