Publications by Econometrics and Free Software
Cluster multiple time series using K-means
I have been recently confronted to the issue of finding similarities among time-series and though about using k-means to cluster them. To illustrate the method, I’ll be using data from the Penn World Tables, readily available in R (inside the {pwt9} package): library(tidyverse) library(lubridate) library(pwt9) library(brotools) First, of all, l...
2335 sym R (4627 sym/10 pcs) 8 img
Cluster multiple time series using K-means
I have been recently confronted to the issue of finding similarities among time-series and though about using k-means to cluster them. To illustrate the method, I’ll be using data from the Penn World Tables, readily available in R (inside the {pwt9} package): library(tidyverse) library(lubridate) library(pwt9) library(brotools) First, of all, l...
2335 sym R (4627 sym/10 pcs) 8 img
Multiple data imputation and explainability
Introduction Imputing missing values is quite an important task, but in my experience, very often, it is performed using very simplistic approaches. The basic approach is to impute missing values for numerical features using the average of each feature, or using the mode for categorical features. There are better ways of imputing missing values, ...
12555 sym R (30783 sym/30 pcs) 18 img
Multiple data imputation and explainability
Introduction Imputing missing values is quite an important task, but in my experience, very often, it is performed using very simplistic approaches. The basic approach is to impute missing values for numerical features using the average of each feature, or using the mode for categorical features. There are better ways of imputing missing values, ...
12555 sym R (30783 sym/30 pcs) 18 img
Intrumental variable regression and machine learning
Intro Just like the question “what’s the difference between machine learning and statistics” has shed a lot of ink (since at least Breiman (2001)), the same question but where statistics is replaced by econometrics has led to a lot of discussion, as well. I like this presentation by Hal Varian from almost 6 years ago. There’s a slide call...
14591 sym R (13242 sym/23 pcs) 14 img
Intrumental variable regression and machine learning
Intro Just like the question “what’s the difference between machine learning and statistics” has shed a lot of ink (since at least Breiman (2001)), the same question but where statistics is replaced by econometrics has led to a lot of discussion, as well. I like this presentation by Hal Varian from almost 6 years ago. There’s a slide call...
14591 sym R (13242 sym/23 pcs) 14 img
Dynamic discrete choice models, reinforcement learning and Harold, part 1
Introduction I want to write about an Econometrica paper written in 1987 (jstor link) by John Rust, currently Professor of Economics at Georgetown University, paper which has been on my mind for the past 10 years or so. Why? Because it is a seminal paper in the econometric literature, but it is quite a bizarre one in some aspects. In this paper, ...
11218 sym R (8893 sym/17 pcs) 10 img 1 tbl
Dynamic discrete choice models, reinforcement learning and Harold, part 1
Introduction I want to write about an Econometrica paper written in 1987 (jstor link) by John Rust, currently Professor of Economics at Georgetown University, paper which has been on my mind for the past 10 years or so. Why? Because it is a seminal paper in the econometric literature, but it is quite a bizarre one in some aspects. In this paper, ...
11217 sym R (8893 sym/17 pcs) 10 img 1 tbl
Dynamic discrete choice models, reinforcement learning and Harold, part 2
In this blog post, I present a paper that has really interested me for a long time. This is part2, where I will briefly present the model of the paper, and try to play around with the data. If you haven’t, I suggest you read part 1 where I provide more context. Rust’s model Welcome to part 2 of this series, which might or might not have a pa...
6432 sym R (5518 sym/11 pcs) 6 img
Dynamic discrete choice models, reinforcement learning and Harold, part 2
In this blog post, I present a paper that has really interested me for a long time. This is part2, where I will briefly present the model of the paper, and try to play around with the data. If you haven’t, I suggest you read part 1 where I provide more context. Rust’s model Welcome to part 2 of this series, which might or might not have a pa...
6432 sym R (5518 sym/11 pcs) 6 img