技能 statsmodels
📊

statsmodels

安全 ⚡ 包含腳本⚙️ 外部命令🌐 網路存取

使用 statsmodels 套用統計模型

也可從以下取得: davila7

使用 OLS、GLM、ARIMA 和離散選擇模型進行嚴謹的統計分析。透過完整的診斷、係數表和殘差分析，獲得可直接發表的結果。

支援: Claude Codex Code(CC)
📊 70 充足
1

下載技能 ZIP

2

在 Claude 中上傳

前往 設定 → 功能 → 技能 → 上傳技能

3

開啟並開始使用

測試它

正在使用「statsmodels」。 Fit an OLS model with robust standard errors and show key diagnostics

預期結果:

  • OLS Regression Results:
  • R-squared: 0.452
  • Coefficient (X1): 2.31 (p < 0.001)
  • Robust HC3 SE: 0.42
  • Heteroskedasticity test (Breusch-Pagan): p = 0.23 (not rejected)
  • Normality test (Jarque-Bera): p = 0.41 (not rejected)

安全審計

安全
v4 • 1/17/2026

Documentation-only skill containing markdown files with Python code examples. Static scanner flagged 434 alerts but all are false positives. The skill contains no executable code - only documentation for the statsmodels statistical library. Scanner misinterpreted markdown backticks as shell commands, statistical terms (HC2, HC3) as C2 indicators, and common patterns as cryptographic algorithms.

8
已掃描檔案
4,331
分析行數
3
發現項
4
審計總數

風險因素

⚡ 包含腳本 (8)
EVALUATION_OUTPUT.json:51 EVALUATION_OUTPUT.json:54 references/discrete_choice.md:94-95 references/discrete_choice.md:328-329 references/discrete_choice.md:448-449 references/time_series.md:114 references/time_series.md:118 references/time_series.md:189
⚙️ 外部命令 (353)
EVALUATION_OUTPUT.json:24 references/discrete_choice.md:28-40 references/discrete_choice.md:40-43 references/discrete_choice.md:43-59 references/discrete_choice.md:59-62 references/discrete_choice.md:62-73 references/discrete_choice.md:73-76 references/discrete_choice.md:76-90 references/discrete_choice.md:90-93 references/discrete_choice.md:93-109 references/discrete_choice.md:109-122 references/discrete_choice.md:122-129 references/discrete_choice.md:129-137 references/discrete_choice.md:137-144 references/discrete_choice.md:144-159 references/discrete_choice.md:159-167 references/discrete_choice.md:167-170 references/discrete_choice.md:170-182 references/discrete_choice.md:182-185 references/discrete_choice.md:185-196 references/discrete_choice.md:196-207 references/discrete_choice.md:207-213 references/discrete_choice.md:213-228 references/discrete_choice.md:228-235 references/discrete_choice.md:235-238 references/discrete_choice.md:238-244 references/discrete_choice.md:244-247 references/discrete_choice.md:247-262 references/discrete_choice.md:262-265 references/discrete_choice.md:265-271 references/discrete_choice.md:271-284 references/discrete_choice.md:284-292 references/discrete_choice.md:292-295 references/discrete_choice.md:295-312 references/discrete_choice.md:312-327 references/discrete_choice.md:327-340 references/discrete_choice.md:340-343 references/discrete_choice.md:343-352 references/discrete_choice.md:352-363 references/discrete_choice.md:363-373 references/discrete_choice.md:373-388 references/discrete_choice.md:388-396 references/discrete_choice.md:396-399 references/discrete_choice.md:399-413 references/discrete_choice.md:413-416 references/discrete_choice.md:416-421 references/discrete_choice.md:421-427 references/discrete_choice.md:427-443 references/discrete_choice.md:443-447 references/discrete_choice.md:447-461 references/discrete_choice.md:461-465 references/discrete_choice.md:465-482 references/discrete_choice.md:482-486 references/discrete_choice.md:486-505 references/discrete_choice.md:505-509 references/discrete_choice.md:509-519 references/discrete_choice.md:519-523 references/discrete_choice.md:523-541 references/discrete_choice.md:541-545 references/discrete_choice.md:545-566 references/discrete_choice.md:566-572 references/discrete_choice.md:572-587 references/discrete_choice.md:587-593 references/discrete_choice.md:593-602 references/discrete_choice.md:602-606 references/discrete_choice.md:606-615 references/discrete_choice.md:615-619 references/discrete_choice.md:619-626 references/discrete_choice.md:626-630 references/discrete_choice.md:630-643 references/discrete_choice.md:643-648 references/glm.md:38-55 references/glm.md:55-58 references/glm.md:58-66 references/glm.md:66-82 references/glm.md:82-97 references/glm.md:97-100 references/glm.md:100-108 references/glm.md:108-119 references/glm.md:119-130 references/glm.md:130-146 references/glm.md:146-154 references/glm.md:154-170 references/glm.md:170-182 references/glm.md:182-197 references/glm.md:197-200 references/glm.md:200-218 references/glm.md:218-223 references/glm.md:223-231 references/glm.md:231-260 references/glm.md:260-277 references/glm.md:277-284 references/glm.md:284-290 references/glm.md:290-305 references/glm.md:305-309 references/glm.md:309-336 references/glm.md:336-340 references/glm.md:340-350 references/glm.md:350-356 references/glm.md:356-371 references/glm.md:371-375 references/glm.md:375-399 references/glm.md:399-403 references/glm.md:403-420 references/glm.md:420-424 references/glm.md:424-445 references/glm.md:445-449 references/glm.md:449-460 references/glm.md:460-464 references/glm.md:464-490 references/glm.md:490-494 references/glm.md:494-523 references/glm.md:523-529 references/glm.md:529-561 references/glm.md:561-565 references/glm.md:565-579 references/glm.md:579-583 references/glm.md:583-593 references/linear_models.md:18-31 references/linear_models.md:31-34 references/linear_models.md:34-44 references/linear_models.md:44-47 references/linear_models.md:47-63 references/linear_models.md:63-66 references/linear_models.md:66-72 references/linear_models.md:72-84 references/linear_models.md:84-94 references/linear_models.md:94-97 references/linear_models.md:97-108 references/linear_models.md:108-120 references/linear_models.md:120-125 references/linear_models.md:125-137 references/linear_models.md:137-144 references/linear_models.md:144-156 references/linear_models.md:156-165 references/linear_models.md:165-177 references/linear_models.md:177-193 references/linear_models.md:193-206 references/linear_models.md:206-227 references/linear_models.md:227-239 references/linear_models.md:239-251 references/linear_models.md:251-257 references/linear_models.md:257-280 references/linear_models.md:280-284 references/linear_models.md:284-306 references/linear_models.md:306-310 references/linear_models.md:310-326 references/linear_models.md:326-330 references/linear_models.md:330-352 references/linear_models.md:352-356 references/linear_models.md:356-379 references/linear_models.md:379-383 references/linear_models.md:383-399 references/linear_models.md:399-405 references/linear_models.md:405-421 references/linear_models.md:421-425 references/linear_models.md:425-427 references/linear_models.md:427-432 references/stats_diagnostics.md:22-31 references/stats_diagnostics.md:31-35 references/stats_diagnostics.md:35-45 references/stats_diagnostics.md:45-49 references/stats_diagnostics.md:49-57 references/stats_diagnostics.md:57-63 references/stats_diagnostics.md:63-72 references/stats_diagnostics.md:72-76 references/stats_diagnostics.md:76-84 references/stats_diagnostics.md:84-88 references/stats_diagnostics.md:88-97 references/stats_diagnostics.md:97-103 references/stats_diagnostics.md:103-115 references/stats_diagnostics.md:115-119 references/stats_diagnostics.md:119-125 references/stats_diagnostics.md:125-129 references/stats_diagnostics.md:129-134 references/stats_diagnostics.md:134-138 references/stats_diagnostics.md:138-143 references/stats_diagnostics.md:143-149 references/stats_diagnostics.md:149-158 references/stats_diagnostics.md:158-162 references/stats_diagnostics.md:162-168 references/stats_diagnostics.md:168-174 references/stats_diagnostics.md:174-191 references/stats_diagnostics.md:191-195 references/stats_diagnostics.md:195-203 references/stats_diagnostics.md:203-211 references/stats_diagnostics.md:211-225 references/stats_diagnostics.md:225-231 references/stats_diagnostics.md:231-249 references/stats_diagnostics.md:249-255 references/stats_diagnostics.md:255-266 references/stats_diagnostics.md:266-272 references/stats_diagnostics.md:272-283 references/stats_diagnostics.md:283-287 references/stats_diagnostics.md:287-298 references/stats_diagnostics.md:298-302 references/stats_diagnostics.md:302-312 references/stats_diagnostics.md:312-320 references/stats_diagnostics.md:320-328 references/stats_diagnostics.md:328-332 references/stats_diagnostics.md:332-341 references/stats_diagnostics.md:341-345 references/stats_diagnostics.md:345-351 references/stats_diagnostics.md:351-357 references/stats_diagnostics.md:357-369 references/stats_diagnostics.md:369-373 references/stats_diagnostics.md:373-381 references/stats_diagnostics.md:381-387 references/stats_diagnostics.md:387-400 references/stats_diagnostics.md:400-404 references/stats_diagnostics.md:404-419 references/stats_diagnostics.md:419-425 references/stats_diagnostics.md:425-433 references/stats_diagnostics.md:433-437 references/stats_diagnostics.md:437-445 references/stats_diagnostics.md:445-449 references/stats_diagnostics.md:449-457 references/stats_diagnostics.md:457-461 references/stats_diagnostics.md:461-467 references/stats_diagnostics.md:467-473 references/stats_diagnostics.md:473-481 references/stats_diagnostics.md:481-485 references/stats_diagnostics.md:485-496 references/stats_diagnostics.md:496-500 references/stats_diagnostics.md:500-508 references/stats_diagnostics.md:508-516 references/stats_diagnostics.md:516-527 references/stats_diagnostics.md:527-531 references/stats_diagnostics.md:531-546 references/stats_diagnostics.md:546-550 references/stats_diagnostics.md:550-560 references/stats_diagnostics.md:560-566 references/stats_diagnostics.md:566-584 references/stats_diagnostics.md:584-590 references/stats_diagnostics.md:590-596 references/stats_diagnostics.md:596-600 references/stats_diagnostics.md:600-608 references/stats_diagnostics.md:608-614 references/stats_diagnostics.md:614-628 references/stats_diagnostics.md:628-632 references/stats_diagnostics.md:632-638 references/stats_diagnostics.md:638-642 references/stats_diagnostics.md:642-656 references/stats_diagnostics.md:656-662 references/stats_diagnostics.md:662-684 references/stats_diagnostics.md:684-688 references/stats_diagnostics.md:688-702 references/stats_diagnostics.md:702-706 references/stats_diagnostics.md:706-731 references/stats_diagnostics.md:731-737 references/stats_diagnostics.md:737-759 references/stats_diagnostics.md:759-763 references/stats_diagnostics.md:763-778 references/stats_diagnostics.md:778-784 references/stats_diagnostics.md:784-795 references/stats_diagnostics.md:795-799 references/stats_diagnostics.md:799-807 references/stats_diagnostics.md:807-813 references/stats_diagnostics.md:813-822 references/stats_diagnostics.md:822-826 references/stats_diagnostics.md:826-833 references/time_series.md:27-36 references/time_series.md:36-39 references/time_series.md:39-43 references/time_series.md:43-46 references/time_series.md:46-50 references/time_series.md:50-66 references/time_series.md:66-74 references/time_series.md:74-79 references/time_series.md:79-102 references/time_series.md:102-105 references/time_series.md:105-127 references/time_series.md:127-130 references/time_series.md:130-149 references/time_series.md:149-164 references/time_series.md:164-174 references/time_series.md:174-177 references/time_series.md:177-184 references/time_series.md:184-187 references/time_series.md:187-197 references/time_series.md:197-213 references/time_series.md:213-232 references/time_series.md:232-235 references/time_series.md:235-245 references/time_series.md:245-248 references/time_series.md:248-258 references/time_series.md:258-275 references/time_series.md:275-293 references/time_series.md:293-296 references/time_series.md:296-311 references/time_series.md:311-314 references/time_series.md:314-325 references/time_series.md:325-328 references/time_series.md:328-333 references/time_series.md:333-344 references/time_series.md:344-354 references/time_series.md:354-366 references/time_series.md:366-371 references/time_series.md:371-374 references/time_series.md:374-385 references/time_series.md:385-391 references/time_series.md:391-406 references/time_series.md:406-410 references/time_series.md:410-436 references/time_series.md:436-440 references/time_series.md:440-456 references/time_series.md:456-462 references/time_series.md:462-484 references/time_series.md:484-488 references/time_series.md:488-505 references/time_series.md:505-509 references/time_series.md:509-518 references/time_series.md:518-522 references/time_series.md:522-548 references/time_series.md:548-554 references/time_series.md:554-571 references/time_series.md:571-575 references/time_series.md:575-598 references/time_series.md:598-602 references/time_series.md:602-620 references/time_series.md:620-624 references/time_series.md:624-643 references/time_series.md:643-651 references/time_series.md:651-658 references/time_series.md:658-664 references/time_series.md:664-675 references/time_series.md:675-681 references/time_series.md:681-690 SKILL.md:33-70 SKILL.md:70-74 SKILL.md:74-105 SKILL.md:105-109 SKILL.md:109-145 SKILL.md:145-149 SKILL.md:149-173 SKILL.md:173-200 SKILL.md:200-228 SKILL.md:228-258 SKILL.md:258-293 SKILL.md:293-336 SKILL.md:336-342 SKILL.md:342-364 SKILL.md:364-370 SKILL.md:370-386 SKILL.md:386-390 SKILL.md:390-405 SKILL.md:405-409 SKILL.md:409-431 SKILL.md:431-437 SKILL.md:437-465 SKILL.md:465-576 SKILL.md:576-585 SKILL.md:585-589
🌐 網路存取 (5)
skill-report.json:6 SKILL.md:608 SKILL.md:609 SKILL.md:610 SKILL.md:611
審計者: claude 查看審計歷史 →

品質評分

45
架構
100
可維護性
81
內容
21
社群
100
安全
87
規範符合性

你能建構什麼

回歸分析

使用穩健標準誤和診斷測試擬合線性和廣義線性模型

計量經濟學建模

使用適當的推論、假設檢定和模型比較來估計因果效應

時間序列預測

建立具有信賴區間和殘差診斷的 ARIMA 模型進行預測

試試這些提示

基本回歸 
使用 statsmodels 擬合具有穩健標準誤的 OLS 回歸模型。顯示係數摘要並檢驗異質變異。
邏輯回歸 
為二元結果建立邏輯回歸模型。計算勝算比、邊際效應和分類指標。
時間序列 
使用 ARIMA 分析此時間序列資料。檢驗平穩性、識別模型階次、擬合模型，並產生 12 步預測和預測區間。
模型比較 
使用 AIC、BIC 和概似比檢定比較幾種模型規格。推薦最佳模型並說明理由。

最佳實務

  • 除非刻意排除截距，否則務必使用 sm.add_constant() 加入常數項
  • 擬合後檢驗模型假設（異質變異、常態性、自相關）
  • 當檢測到假設違規時，使用穩健標準誤（HC0-HC3）
  • 報告推論時一併提供信賴區間和點估計

避免

  • 忘記加入常數項會導致截距估計偏誤
  • 忽略異質變異會使推論失效
  • 對二元結果使用 OLS 會產生不正確的機率
  • 解釋前未檢查有影響力的觀測值

常見問題

OLS 和 GLM 有什麼區別？
OLS 假設誤差項呈常態分配且變異數固定。GLM 透過使用不同的分佈和連結函式，將此擴展到非常態結果。
如何選擇 Logit 和 Probit？
兩者對二元結果的效果相似。Logit 的尾部較厚。對於較易於解釋（勝算比）使用 Logit，或因理論原因使用 Probit。
ARIMA 應該使用什麼階數？
使用 ACF/PACF 圖來識別 p（AR 階數）和 q（MA 階數）。使用 ADF 檢驗平穩性以確定 d（差分階數）。
何時應該使用穩健標準誤？
當存在異質變異或資料中有叢集導致相關性時，使用穩健標準誤。
如何比較嵌套模型？
使用概似比檢定。對於非嵌套模型，比較 AIC 或 BIC 值。較低的 AIC/BIC 表示較好的配適。
擬合後應該執行哪些診斷檢定？
檢驗異質變異（Breusch-Pagan）、常態性（Jarque-Bera）、自相關（Durbin-Watson），並檢查有影響力的觀測值（Cook's distance）。

開發者詳情

作者

K-Dense-AI

授權

BSD-3-Clause license

儲存庫

https://github.com/K-Dense-AI/claude-scientific-skills/tree/main/scientific-skills/statsmodels

引用

main

檔案結構

📁 references/

📄 discrete_choice.md

📄 glm.md

📄 linear_models.md

📄 stats_diagnostics.md

📄 time_series.md

📄 EVALUATION_OUTPUT.json

📄 SKILL.md