econometrics-julia

本条目面向分析数据与指标归因场景,提供“econometrics-julia”技能的目录入口。公开标签:learning、Julia、计量经济学、结构估计。请通过固定上游版本核对实际功能、依赖与权限;此页为 Skillπ 独立编写的用途介绍和原站链接,不代表运行安全保证。原文与附件依各自许可开放。

仓库
brycewang-stanford/Auto-Empirical-Research-Skills
仓库最近推送
Stars
5,023
Forks
578
Issues / PRs
0

安装方式

先用 Prompt 核对来源,也可切换到安装命令或下载可用的安装包。

安装方式

用 AI 助手安装

复制提示词,让助手先核对固定版本和依赖。

查看 Prompt收起 Prompt
先检查当前 Agent 是否已安装 SkillPI Skill Finder。如果尚未安装,请按官方仓库的说明安装并启用:https://github.com/Findskills/skillpi。然后使用 Skill Finder 打开此 SkillPI 详情页并定位 "econometrics-julia":https://skillpi.com/zh/skills/deepskill-bede8b06711886f1c00b34eb0430afb40c6b834f278f36ce06c2f39a7db59f7a。安装前核对上游来源和版本:https://github.com/brycewang-stanford/Auto-Empirical-Research-Skills/blob/bfa0a6e5a843babbf45a1ef1663fd5dc4f07c0ba/skills/14-luischanci-claude-code-research-starter/dot-claude/skills/econometrics-julia/SKILL.md。来源不一致时停止并说明差异。核对通过后,用 Skill Finder 安装目标 Skill,并报告安装路径和来源版本。

此命令从 GitHub 来源仓库获取指定技能。运行前请先查看源文件。

运行工具
npx skills add https://github.com/brycewang-stanford/Auto-Empirical-Research-Skills/tree/bfa0a6e5a843babbf45a1ef1663fd5dc4f07c0ba/skills/14-luischanci-claude-code-research-starter/dot-claude/skills/econometrics-julia --skill econometrics-julia
许可协议: —

来源与配置

查看 GitHub 源码

brycewang-stanford/Auto-Empirical-Research-Skills固定提交 bfa0a6e5a843babbf45a1ef1663fd5dc4f07c0ba

配置方式待核实

请查看原始来源中的适用工具、依赖和权限要求。

安装实测待补充

本条目面向分析数据与指标归因场景,提供“econometrics-julia”技能的目录入口。公开标签:learning、Julia、计量经济学、结构估计。请通过固定上游版本核对实际功能、依赖与权限;此页为 Skillπ 独立编写的用途介绍和原站链接,不代表运行安全保证。原文与附件依各自许可开放。

↓ 0 下载 ↗ 0 分享点击

文件浏览

在原站浏览文件 ↗

原始 SKILL.md

打开原文 ↗
---
name: econometrics-julia
description: Julia-based econometric and structural estimation for computationally intensive tasks. Use for structural models, maximum likelihood, GMM, numerical optimization, simulations, and high-performance computing. Covers DataFrames.jl, FixedEffectModels.jl, Optim.jl, and performance optimization.
---

# Julia Econometrics & Structural Estimation

## Core Packages
```julia
using DataFrames, CSV       # Data handling
using FixedEffectModels     # Panel regressions
using Optim                 # Optimization
using Distributions         # Probability distributions
using LinearAlgebra         # Matrix operations
using ForwardDiff           # Automatic differentiation
using Statistics, StatsBase
```

## Data Handling (DataFrames.jl)
```julia
# Read data
df = CSV.read("data/raw/file.csv", DataFrame)

# Basic operations
transform!(df, :x => (x -> x .* 100) => :x_scaled)
subset!(df, :year => y -> y .>= 2000)

# Groupby operations
gdf = groupby(df, :id)
combine(gdf, :y => mean => :mean_y, nrow => :n)

# Efficient joins
leftjoin!(df, other_df, on = [:id, :year])
```

## Panel Regressions (FixedEffectModels.jl)
```julia
# Two-way fixed effects
est = reg(df, @formula(y ~ treatment + fe(id) + fe(year)), 
          Vcov.cluster(:state))

# IV estimation
est_iv = reg(df, @formula(y ~ (endog ~ instrument) + fe(id) + fe(year)))

# Extract results
coef(est)        # Coefficients
vcov(est)        # Variance-covariance matrix
nobs(est)        # Observations
r2(est)          # R-squared
```

## Performance Optimization

### Memory Efficiency
```julia
# Use views instead of copies
@views y_subset = df.y[1:100]

# Preallocate arrays
result = zeros(n, m)
@inbounds for i in 1:n
    result[i, :] .= compute_row(i)
end

# Use StaticArrays for small fixed-size arrays
using StaticArrays
params = SVector{3, Float64}(1.0, 2.0, 3.0)
```

### Broadcasting and Vectorization
```julia
# Good: fused broadcasting
@. result = x^2 + y^2 + z  # Single pass

# Avoid: creates intermediates
result = x.^2 .+ y.^2 .+ z  # Multiple allocations
```

### Type Stability
```julia
# Always annotate function return types for critical code
function likelihood(θ::Vector{Float64}, data::Matrix{Float64})::Float64
    # Implementation
end

# Use concrete types in structs
struct ModelParams{T<:Real}
    α::T
    β::T
    σ::T
end
```

## Structural Estimation

### Maximum Likelihood
```julia
function log_likelihood(θ, data)
    α, β, σ = θ
    y, X = data[:, 1], data[:, 2:end]
    
    resid = y .- X * [α, β]
    ll = -0.5 * length(y) * log(2π * σ^2) - 
         sum(resid.^2) / (2σ^2)
    return -ll  # Minimize negative log-likelihood
end

# Optimize
θ0 = [0.0, 0.0, 1.0]
result = optimize(θ -> log_likelihood(θ, data), θ0, LBFGS(),
                  autodiff = :forward)

θ_hat = Optim.minimizer(result)
```

### Generalized Method of Moments (GMM)
```julia
function gmm_moments(θ, data, Z)
    α, β = θ
    y, X = data[:, 1], data[:, 2:end]
    resid = y .- X * [α, β]
    
    # Moment conditions: E[Z'ε] = 0
    moments = Z' * resid / size(Z, 1)
    return moments
end

function gmm_objective(θ, data, Z, W)
    g = gmm_moments(θ, data, Z)
    return g' * W * g
end

# Two-step GMM
W1 = I(size(Z, 2))  # First step: identity
θ1 = optimize(θ -> gmm_objective(θ, data, Z, W1), θ0).minimizer

# Optimal weighting matrix
Ω = cov(Z .* (data[:, 1] .- data[:, 2:end] * θ1))
W2 = inv(Ω)
θ2 = optimize(θ -> gmm_objective(θ, data, Z, W2), θ1).minimizer
```

### Simulation-Based Estimation (MSM/Indirect Inference)
```julia
function simulate_model(θ, N, S; seed=12345)
    Random.seed!(seed)
    α, β, σ = θ
    
    # Simulate S datasets of size N
    sims = Matrix{Float64}(undef, N, S)
    for s in 1:S
        X = randn(N)
        ε = σ * randn(N)
        sims[:, s] = α .+ β .* X .+ ε
    end
    return sims
end

function msm_objective(θ, data_moments, N, S)
    sims = simulate_model(θ, N, S)
    sim_moments = mean([compute_moments(sims[:, s]) for s in 1:S])
    diff = data_moments - sim_moments
    return diff' * diff
end
```

## Numerical Methods

### Root Finding (for equilibrium models)
```julia
using NLsolve

function excess_demand(p, params)
    # Market clearing conditions
    return demand(p, params) - supply(p, params)
end

result = nlsolve(p -> excess_demand(p, params), p0)
p_eq = result.zero
```

### Integration (for expected values)
```julia
using QuadGK

# One-dimensional
expected_value, err = quadgk(x -> x * pdf(Normal(μ, σ), x), -Inf, Inf)

# Multi-dimensional (Monte Carlo or sparse grids)
using HCubature
integral, err = hcubature(f, lower_bounds, upper_bounds)
```

## Parallel Computing
```julia
using Distributed
addprocs(4)

@everywhere using SharedArrays

# Parallel bootstrap
@distributed for b in 1:B
    bootstrap_sample = sample(1:N, N, replace=true)
    θ_boot[b] = estimate(data[bootstrap_sample, :])
end

# Or use Threads for shared memory
Threads.@threads for i in 1:N
    result[i] = compute(i)
end
```

## Standard Errors

### Bootstrap
```julia
function bootstrap_se(estimate_fn, data, B=1000)
    N = size(data, 1)
    θ_boot = [estimate_fn(data[sample(1:N, N, replace=true), :]) 
              for _ in 1:B]
    return std(θ_boot)
end
```

### Delta Method
```julia
using ForwardDiff

function delta_method_se(g, θ, Σ)
    # g: transformation function
    # θ: parameter estimates
    # Σ: variance-covariance of θ
    ∇g = ForwardDiff.gradient(g, θ)
    return sqrt(∇g' * Σ * ∇g)
end
```

## Best Practices
- Use `@time` and `@btime` (from BenchmarkTools) to profile
- Check type stability with `@code_warntype`
- Use `const` for global constants
- Prefer column-major iteration (Julia is column-major)

版本历史

  1. v2026.07 GitHub ↗
    原版更新说明

    ## Highlights - **First tagged release.** Every gate is green: `make check` (validate + 183 unit tests + eval-harness + numeric benchmarks), validate-catalog, quality-evals, and OpenSSF Scorecard on `main`. - **Rigor double loop at 13 method families.** Each family closes taxonomy tag → eval scenario (does the agent avoid the method's known traps?) → deterministic numeric benchmark (does it recover the true effect?). New this cycle: heterogeneous treatment effects (CATE) and quantile treatment effects (QTE). - **Paper-WorkFlow flagship at 29/29 executable rigor gates**, now merged to its `main` with a drift-gated README badge — the credibility artifact is checked by CI, so it can never silently go stale ([RIGOR.md](https://github.com/brycewang-stanford/Paper-WorkFlow/blob/main/RIGOR.md)). - **1,150 vendored skills / 69 collections**, all license-audited, security-scanned, tracked in a machine-readable catalog with CI-enforced freshness, and importable as a single whole-repo router skill. - **Hardened weekly upstream syncs**: synced trees must pass the full gate before merge; this cycle both sync PRs were verified, patched (StatsPAI's SkillOpt execution-gate card restored), and merged. - **First-party plugin marketplace** (Python / R / Stata empirical-analysis plugins + guide) shipped in the previous cycle, now under a tagged version. ## Known Follow-Ups - Ecosystem benchmark comparison (AERS vs econometrics agents on the shared benchmark suite) — [PLAN-2026-07 week 3](https://github.com/brycewang-stanford/Auto-Empirical-Research-Skills/blob/main/docs/PLAN-2026-07.md). - Decomposition (Oaxaca–Blinder) and shift-share (Bartik) method families (weeks 3–4). - GitHub Pages catalog site, README demo screenshots, five-language README consistency script (week 4). ## Catalog - Top-level collections: **69** - `SKILL.md` files: **1150** - Cataloged agent tools: **335** ([`tools/CATALOG.md`](../tools/CATALOG.md)) ## Methodological rigor - Method families in the coverage map: **13** ([`RIGOR_COVERAGE.md`](RIGOR_COVERAGE.md)) - Eval scenarios: **26** (138 rubric items, 122 auto-checkable) ([`../eval-harness/`](../eval-harness/README.md)) - Benchmark tasks: **13** (70 gold checks, 58 required) ([`../benchmark/`](../benchmark/README.md)) ### Method families (skills tagged) - `iv`: 27 - `rdd`: 25 - `panel-fe`: 19 - `staggered-did`: 17 - `bayesian`: 13 - `did`: 10 - `event-study`: 10 - `matching`: 10 - `synthetic-control`: 10 - `cate`: 8 - `dml`: 8 - `survival`: 3 - `quantile`: 2 ### Eval scenarios by category - `causal-identification`: 14 - `citation-hygiene`: 1 - `reproducibility`: 2 - `research-integrity`: 1 - `runtime-safety`: 1 - `statistical-validity`: 2 - `writing-compliance`: 3 - `writing-style`: 2 ## License buckets - AGPL-3.0: 1 - Apache-2.0: 1 - CC-BY-NC-4.0 (non-commercial): 1 - CC-BY-SA-4.0: 1 - CC-BY-SA-4.0 (repository default): 4 - GPL-3.0: 1 - MIT: 35 - MIT Non-Commercial: 1 - Mixed (per-skill; see folder): 2 - UNKNOWN - check upstream: 22 ## Quality gates - `make check` — catalog/provenance/audit/eval freshness, repo hygiene, Paper-WorkFlow gates, stdlib unit tests, eval-harness lint + graded smoke, and the strict numeric benchmark (reference-gold drift fails the build). - External links and OpenSSF Scorecard run on their scheduled workflows.

当前收录版本

0.0.1

核对来源版本 ↗

质量评测

满分 100。前四项来自 DeepSkill 收录的质量报告。

场景价值92/100
评分依据

提供完整的 Julia 计量经济学与结构估计工具链,涵盖从数据处理到高级估计方法的端到端解决方案

触发精准88/100
评分依据

description 清晰界定适用范围,涵盖 structural models、maximum likelihood、GMM、numerical optimization 等核心能力

结构规范95/100
评分依据

结构清晰,分模块组织,每部分含可执行代码示例,便于用户直接复用

执行可靠90/100
评分依据

代码示例参数正确,工具来源明确(DataFrames.jl、FixedEffectModels.jl、Optim.jl 等主流包)

证据可验 (前四项均值)91/100
计算方式

取前四项原始分值的算术平均,四舍五入取整。可通过收录报告核对各项评分与说明,运行记录需单独查证。

权限与使用前检查

根据当前主文档自动整理,命中内容可能是示例或防护说明;不是实际权限声明,也未覆盖附件、外部依赖或运行行为。

未从主文档识别到上述常见操作,不能据此判断无需权限或没有风险。

本站展示提示不会授予权限。涉及文件改动、密钥、外部发送或费用时,仍由你使用的工具按需确认。

安全评级与访问范围

risk · 静态证据核验,不代表运行安全保证。

使用前请由宿主核对并授权实际文件读写、外部服务和联网目的地;本站的核验不会授予任何运行权限。

来源与使用声明 SkillPi 的部分技能资料来自公开来源或创作者提交。第三方内容的权利归各自权利人所有。安装或运行前,请查看原站的许可、权限要求和使用风险。

来源与使用

请按详情页的作者、原始来源和许可证核对使用条件,运行前检查技能需要的权限。目录收录不等于已验证安装或安全性。