update test dataset

.gitattributes

@@ -57,3 +57,7 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 # Video files - compressed
 *.mp4 filter=lfs diff=lfs merge=lfs -text
 *.webm filter=lfs diff=lfs merge=lfs -text
+problem_set/ps1_24S_cleaned.dta filter=lfs diff=lfs merge=lfs -text
+published_paper/DID_Canal/cao_chen_cleaned_3.dta filter=lfs diff=lfs merge=lfs -text
+published_paper/DID_BBB/macro_workfile.dta filter=lfs diff=lfs merge=lfs -text
+published_paper/IV_Birth_Quarter/Birth_Quarter_1980_30_39.dta filter=lfs diff=lfs merge=lfs -text

README.md

@@ -7,10 +7,9 @@ tags:
 
 # Data Card for Econometric AI Agent Testset
 
-## Dataset Description
-- **Repository:** https://github.com/HKU-Business-AI-Center/Econometrics-Agent
-- **Paper:** https://arxiv.org/abs/2506.00856
+### Dataset Summary
 
+This sample test dataset pertains to the study titled [Can AI Master Econometrics? Evidence from Econometrics AI Agent on Expert-Level Tasks](https://arxiv.org/abs/2506.00856). Our goal is to establish a standardized benchmark to evaluate the ability of artificial intelligence models or agents in executing econometric tasks. We invite contributions to expand this econometric task repository and to assess the performance of various large language models (LLMs) and AI agents. If you use our dataset, please kindly cite the original source.
 
 ## Citation Information
 ```bibtex
@@ -24,3 +23,10 @@ tags:
       url={https://arxiv.org/abs/2506.00856}, 
 }}
 ```
+
+## Dataset Description
+- **Repository:** https://github.com/HKU-Business-AI-Center/Econometrics-Agent
+- **Paper:** https://arxiv.org/abs/2506.00856
+
+
+

problem_set/ps1_24S_cleaned.dta

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c75b39aebb5077bdd7b1b35a92b937e77f511ecce8c3b6d0961bad59e521c408
+size 15059809

problem_set/regression_set.csv

@@ -0,0 +1,6 @@
+,y,x,control_variables,data_source,method,other_requirements,answer,level,journal,article,notation,tags
+0,dbrwt,tobacco,"rectype, csex, dmar, pldel3, pre4000, preterm, alcohol, dmage, dmeduc, dlivord, monpre, nprevist, dplural, birattnd, cntocpop, ormoth, mrace3, adequacy, delmeth5",ps1_24S_cleaned.dta,OLS,"birattnd, cntocpop, ormoth, mrace3, adequacy, delmeth5 are multi-class categorical variables","{'coefficient': -212.9425,'standard_error': 4.579,'p_value': 0.0}",,,,ps1_B4,
+1,dbrwt,tobacco,"rectype, csex, dmar, pldel3, pre4000, preterm, alcohol, dmage, dmeduc, dlivord, monpre, nprevist, dplural, birattnd, cntocpop, ormoth, mrace3, adequacy, delmeth5",ps1_24S_cleaned.dta,propensity score regression,"birattnd, cntocpop, ormoth, mrace3, adequacy, delmeth5 are multi-class categorical variables","{'coefficient': -209.835,'standard_error': 5.044,'p_value': 0.0}",,,,ps1_C1 & ps1_C2,
+2,log_fatality_rate,primary,,ps2_24S_cleaned.dta,panel OLS,"need to constuct log_fatality_rate as the dependent variable, by taking the direct logarithm of fatality_rate. Compute the robust standard errors ","{'coefficient': -0.1429,'standard_error': 0.0284,'p_value': 0.0}",,,,ps2_A2_1,"data processing, robust standard error"
+3,log_fatality_rate,primary,,ps2_24S_cleaned.dta,panel OLS,"need to constuct log_fatality_rate as the dependent variable, by taking the direct logarithm of fatality_rate. add state fixed effect and cluster standard error by state & year","{'coefficient': -0.2055,'standard_error': 0.0331,'p_value': 0.0}",,,,ps2_B1_1,"data processing, cluster standard error, fixed effect"
+4,log_fatality_rate,primary,,ps2_24S_cleaned.dta,Staggered DID Event Study,"for the event study setting, set the see-back length as 4 and see-forward length as 3. Need to constuct log_fatality_rate as the dependent variable, by taking the direct logarithm of fatality_rate. Add state fixed effect, year fixed effect and cluster standard error by state. Output the result of coefficient towards ""Lag_D1"".","{'coefficient': -0.0430,'standard_error': 0.0176,'p_value': 0.0147}",,,,ps2_B2,"data processing, cluster standard error, fixed effect"

published_paper/DID_BBB/macro_workfile.dta

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e05281f9ecfc5674a8d4bc79dcbee65905bb3d8e61bbba55ed2ea38fdedbedbd
+size 860558

published_paper/DID_Canal/cao_chen_cleaned_3.dta

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:859d0629f9737ff9ba354f8bb53cb8089020c18343f79f4350a5397a69725212
+size 19932145

published_paper/IV_Birth_Quarter/Birth_Quarter_1980_30_39.dta

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c7182c1b1909a75eeabfc8ecda4d6fe6612d23f414c27f75ec0c2ad61752c089
+size 8249800

published_paper/regression_testset.csv

@@ -0,0 +1,6 @@
+,y,x,control_variables,data_source,method,other_requirements,answer,level,journal,article,notation,tags
+9,lwklywge,educ,"race, smsa, married, yob",IV_Birth_Quarter/Birth_Quarter_1980_30_39.dta,IV 2SLS Regression,"yob is a multi-class categorical variable, and should first be constructed into dummies as the control variables; qob is also a multi-class categorical variable, construct into dummies and use these dummies as instrument variables","{'coefficient': 0.0989901,'standard_error': 0.0206919,'p_value': 0.000}",,Quarterly Journal of Economics,Does Compulsory School Attendance Affect Schooling and Earnings?,IV_Birth_Quarter,data processing
+10,loggdp,risk,"latitude, asia, africa, other",IV_Colonial_Origins/Colonial_Origins_Reply.dta,IV 2SLS Regression,use logmort0 as the instrument variable; use robust standard error,"{'coefficient': 1.073909,'standard_error': 0.4835807,'p_value': 0.026}",,American Economic Review,The Colonial Origins of Comparative Development: An Empirical Investigation,IV_Colonial_Origins,robust standard error
+11,lngini,_intra,,DID_BBB/macro_workfile.dta,Staggered DID Event Study,"need to construct lngini by taking the direct logarithm of gini; for the event study setting, set the see-back length as 10 and see-forward length as 15. Add state fixed effect towards statefip, year fixed effect towards wrkyr, and cluster standard error by state. Output the result of coefficient towards the parameter ""Lag_D1"".","{'coefficient': -0.0005764,'standard_error': 0.0076168,'p_value': 0.940}",,Journal of Finance,Big Bad Banks? The Winners and Losers from Bank Deregulation in the United States,DID_BBB,"data processing, cluster standard error, fixed effect"
+12,rebel,canal,,DID_Canal/cao_chen_cleaned_3.dta,Static DID Regression,"canal denotes if an entity receives treatment, and post denote if an entity in the treatment group has received treatement;add entity fixed effect towards county, time fixed effect towards year, and cluster standard error by county. Output the result of coefficient towards the interaction term parameter ""treatment_group_treated"".","{'coefficient': 0.0380143,'standard_error': 0.016621,'p_value': 0.023}",,American Economic Review,"Rebel on the Canal: Disrupted Trade Access and Social Conflict in China, 1650-1911",DID_Canal,fixed effect
+13,fte,treated,,DID_MinWage/cardkrueger1994_cleaned.dta,Static DID Regression,"treated denotes if an entity receives treatment, and t denote if an entity in the treatment group has received treatement;add entity fixed effect towards id, time fixed effect towards t, and cluster standard error by id. Output the result of coefficient towards the interaction term parameter ""treatment_group_treated"".","{'coefficient': 2.986941,'standard_error': 1.33263,'p_value': 0.026}",,American Economic Review,Minimum Wages and Employment: A Case Study of the Fast Food Industry,DID_MinWage,"fixed effect, cluster standard error"