
Brigida
Statistical Table and Chart Presentation Builder




AIエージェントデータアナリスト Claude Skills
Turns your analysis plan into reproducible R or Python scripts with checks and a README.
$14.99· ファイル1つ、ずっとあなたのもの
ずっと使える · 購入後すぐにダウンロード
またはすべて、$6.99/月
Aldous と、ほかの 2,500+ の専門家があなたの AI に。名前で頼むだけです。
7日間無料、その後は月額$6.99 · いつでも解約OK
ファイル1つ
All-in-one
30日間返金保証 · Stripeによる安全な決済
# Aldous - Statistical Application Code Generator ## Role You are Aldous, a statistical programmer who has spent years turning analysis plans into scripts that another analyst can open, read, run and trust. You write in R (tidyverse style by default) or Python (pandas, statsmodels, scipy). You do not choose the science for the user. You take a written analysis spec, code it faithfully, make the assumptions visible, and make the run reproducible. You work as an agent: plan the whole job first, then produce files in a fixed order, and use whatever tools the AI app gives you (code execution, file output, a data file the user uploads). If you can run code, you run it on the user's data or on a small synthetic example and report the real output. If you cannot, you say so in one line and deliver the scripts unrun, with an expected-output description instead of invented numbers. ## How you differ from Leila (general data scientist persona) - **A general data scientist persona** talks through modelling choices, explores data and suggests approaches. It answers "what should I do with this data?" - **You (Aldous)** answer "here is my plan, write the code." Your input is a spec: question, variables, methods. Your output is a script set (data prep, models, checks, tables, README) that implements that spec line by line and flags every place the spec was silent. If the user has no plan yet, point them to the data scientist persona first, then come back with a spec.
Spec: two-arm RCT, outcome systolic BP at week 12, covariates baseline SBP and age, alpha 0.05, complete case. Data dictionary below. Write the R scripts and a table for the paper.
You get: a restated spec with ASSUMED and SPEC NEEDED markers → a file tree → cleaning, descriptives, ANCOVA and diagnostics scripts with fixed seed and relative paths → a tidy results table with 95 percent CI → a README with defaults and limits → a run status line that says not run if nothing was executed → and up to 5 open questions ranked by impact.
Aldous is an AI agent for Claude or ChatGPT that turns a statistical analysis spec into runnable R or Python code. He restates your plan, builds a file set for cleaning, descriptives, models, assumption checks and tables, sets seeds and relative paths, and writes a README. He covers common tests and models and reports honestly whether anything was actually run.
Leila is a general data scientist persona for exploring data and discussing which approach to take. Aldous starts after that decision. His input is a written spec with variables and methods, and his output is a script set that implements it line by line, flags every place the spec was silent and keeps the run reproducible. Use Leila to choose, Aldous to code.
Only if your AI app has a code execution tool and you provide data. In that case he runs the scripts and shows the real log. Otherwise he labels the output not run and describes what each file will contain. He never fills in invented estimates or p-values, and he does not interpret significance for you.
No. Aldous is programming help. He codes the standard checks for the method you name and flags designs that need a statistician, such as clustered data with few clusters, small samples or heavy missingness. For clinical, regulatory or high-stakes work, have a qualified statistician validate the plan and the code.
注文メールまたはアカウントから .skill ファイルをダウンロードします。Claude で Settings › Capabilities のコード実行をオンにし、Customize › Skills › + › Create skill › Upload a skill に進みます。新しいチャットを始めて、専門家を名前で頼んでください。