Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
P 値と有意差/分散分析 / P-value, Significant Difference ...
Search
Kenji Saito
PRO
January 03, 2025
Technology
160
0
Share
Embed
Copy iframe code
Copy JS code
Copy link
Start on current slide
P 値と有意差/分散分析 / P-value, Significant Difference and Analysis of Variance
早稲田大学大学院経営管理研究科「企業データ分析」2024 冬の第9-10回で使用したスライドです。
Kenji Saito
PRO
January 03, 2025
More Decks by Kenji Saito
See All by Kenji Saito
戦略と安全保障 / Strategy and Security
ks91
PRO
0
22
ローカル LLM とコンピューティングの脱領土化 / Local LLMs and Deterritorialization of Computing
ks91
PRO
0
13
サイバー戦の現在と未来 / The Present and Future of Cyber Warfare
ks91
PRO
0
71
金融の未来を考える / Thinking About the Future of Finance
ks91
PRO
0
150
AI Agentification: Current State and the Changes Ahead
ks91
PRO
0
40
サイバー攻撃・防御とサイバー法 / Cyber Attacks, Defense, and Cyber Law
ks91
PRO
0
100
AI・ロボティクスと自動化社会 / AI, Robotics, and the Automated Society
ks91
PRO
0
150
ロボティクスの技術 / Robotics Technology
ks91
PRO
0
170
インシデントレスポンス演習 I / Incident Response Exercise I
ks91
PRO
0
140
Other Decks in Technology
See All in Technology
アラート調査向けAIエージェントの本番導入とその後/AI Agents for Alert Investigation: Production Deployment and After
taddy_919
1
330
SRE歴2ヶ月でも開発6年の知見を活かして、チームで止まっていた環境改善を前に進めた話
a_ono
0
180
Text-to-SQLをAgentCoreで実現し、生成されるSQLの精度を定量的に評価する
yakumo
2
470
しぶいSRE: サーバから見えない障害にどう向き合うか。ラストワンマイルのデバッグ実践 / Shibui SRE
kanny
0
210
Claude Codeとハーネスについて考えてみる
oikon48
16
7.3k
事業価値を⽣み出すSREへ SREが担うべき意思決定の5層
kenta_hi
0
130
“ID沼入口” - 基本とセキュリティから始める、考え続けるためのID管理技術勉強会 告知&イントロ
ritou
0
380
AIに障害切り分けを全部やってもらった。 。 。 。
estie
0
340
AIペネトレーションテスト・ セキュリティ検証「AgenticSec」紹介資料
laysakura
2
8k
GitHub Copilot運用のリアル ~AI Credit時代にどう向き合うか~
takafumisu2uk1
0
620
Baseline対応のDOMの型定義を作った
uhyo
2
650
次世代ランサムウェア対策の考察 / 20260704 Mitsutoshi Matsuo
shift_evolve
PRO
5
1.7k
Featured
See All Featured
Stewardship and Sustainability of Urban and Community Forests
pwiseman
0
240
技術選定の審美眼(2025年版) / Understanding the Spiral of Technologies 2025 edition
twada
PRO
118
120k
Understanding Cognitive Biases in Performance Measurement
bluesmoon
32
3k
Designing for Performance
lara
611
70k
Optimizing for Happiness
mojombo
378
71k
個人開発の失敗を避けるイケてる考え方 / tips for indie hackers
panda_program
123
22k
Context Engineering - Making Every Token Count
addyosmani
9
1k
SEO in 2025: How to Prepare for the Future of Search
ipullrank
3
3.6k
The Pragmatic Product Professional
lauravandoore
37
7.3k
Evolution of real-time – Irina Nazarova, EuRuKo, 2024
irinanazarova
9
1.4k
Building a Scalable Design System with Sketch
lauravandoore
463
34k
Building a A Zero-Code AI SEO Workflow
portentint
PRO
0
620
Transcript
Corporate data analysis — generated by Stable Diffusion XL v1.0
2024 9-10 P (WBS) 2024 9-10 P — 2025-01-06 – p.1/33
https://speakerdeck.com/ks91/collections/corporate-data-analysis-2024-winter 2024 9-10 P — 2025-01-06 – p.2/33
( ) 1 12 2 • 2 12 2 (B
A ) • 3 12 9 • 4 12 9 • 5 12 16 • 6 12 16 t • 7 12 23 2 ( ) t • 8 12 23 2 ( ) t • 9 1 6 P • 10 1 6 • 11 1 20 12 1 20 13 1 27 14 1 27 W-IOI 2024 9-10 P — 2025-01-06 – p.3/33
( 20 25 ) 1 (20 ) • 2 R
( 55 ) • 3 (32 ) • 4 (14 ) • 5 ( Git) (22 ) • 6 ( ) (24 ) • 7 (1) (25 ) • 8 (2) (25 ) • 9 R ( ) (1) — Welch (17 ) • 10 R ( ) (2) — (21 ) • 11 R ( ) (1) — (15 ) • 12 R ( ) (2) — (19 ) • 13 GPT-4 (19 ) • 14 GPT-4 (29 ) • 15 ( ) LaTeX Overleaf (40 ) • 8 (12/16 ) / (2 ) OK / 2024 9-10 P — 2025-01-06 – p.4/33
( Student µ 95% ) 7 2 t ( t
) 2 ( ) 2 d ( ) ← [ 3] σd 2 t 8 2 t ( t ) 2 ( ) ( ) ← [ 4] σ 2 t 2024 9-10 P — 2025-01-06 – p.5/33
2 2 t 1 9 P P 10 H0 HA
k, N, ¯ ¯ x σ2 ( )MSwithin ( )MSbetween MStotal F F 2024 9-10 P — 2025-01-06 – p.6/33
2024 9-10 P — 2025-01-06 – p.7/33
4. t (1) 2 t (2) 2 t (3) 2025
1 2 ( ) 23:59 JST ( ) Waseda Moodle (Q & A ) (1)(2) Discord 2024 9-10 P — 2025-01-06 – p.8/33
. . . . . . 17 14 (1/3( )
) ( ) → 14 ( ) ( ) → 6 → 3 ( ) → 5 ( ) ( OK) 2 t . . . . . . / . . . ( ) 2024 9-10 P — 2025-01-06 – p.9/33
t t ⇒ ( ) A A xA 2 B
B xB 2 df . . . ⇒ t σ z0.05 . . . ⇒ ( ) t 2024 9-10 P — 2025-01-06 – p.10/33
N (1/2) 2 t 2 2 “ ” 1. 1
2 2. 3. - 2 (n − 1) 4. ÷ ÷ t 5. t (n − 1) t t ⇒ . . . 0 ( ) 2024 9-10 P — 2025-01-06 – p.11/33
N (2/2) 2 t 2 2 1 2 1 2
2 “ ” 1. 2 1 2 2. 3. ( -2) 4. t 1÷ 2 t 5. t (n1 + n2 − 2) t t ⇒ 2024 9-10 P — 2025-01-06 – p.12/33
M ( ) [ 2 t ] 1Day 1Day 1Day
⇒ 2024 9-10 P — 2025-01-06 – p.13/33
K ⇒ . . . 2024 9-10 P — 2025-01-06
– p.14/33
2 t d : µd 0 ( 2 ) :
(1) d d, sd , n, df (2) |d| sd n |t| (3) t0.05 (df) < |t| ( ) R > t.test(sample2, sample1, paired=T) 2024 9-10 P — 2025-01-06 – p.15/33
2 t ( ) 10 ( ) ( ) (
) ( ) ( ) d ( ) d, ( ) sd , ( ) n, ( ) df ( ) t ( ) t ( ) d ( ) sd ( ) n ( ) t df 5% ( ) ( ) ( ) ( ) ( ) 2024 9-10 P — 2025-01-06 – p.16/33
2 t xA xB : µA − µB 0 (
2 ) : (1) xA − xB , sp , nA nB , df (2) |xA − xB | sp nA nB |t| (3) t0.05 (df) < |t| ( ) R > t.test(sample2, sample1, var.equal=T) 2024 9-10 P — 2025-01-06 – p.17/33
2 t ( ) ( ) ( ) ( )
( ) ( ) ( ( ) A B ( ) ) ( ) xA − xB , A B ( ) ( ) sp , ( ) nA nB , ( ) df = nA + nB − 2 ( ) t ( ) t ( ) xA − xB ( ) sp ( ) nA ,nB ( ) t df 5% ( ) ( ) ( ) ( ) ( ) ( ) 2024 9-10 P — 2025-01-06 – p.18/33
K 2 t ( ) ⇒ 2 2024 9-10 P
— 2025-01-06 – p.19/33
N ⇒ (σ) ( σ √n ) ( ) p.121
(standard error) (p.121) (sampling distribution) (p.120) (p.120) ( : ) 2024 9-10 P — 2025-01-06 – p.20/33
K ⇒ . . . AI ( ) . .
. ^^; ( ) 2024 9-10 P — 2025-01-06 – p.21/33
H t 2 Student t t 1 sin(α + β)
= sinαcosβ + cosαsinβ . . . ⇒ 2024 9-10 P — 2025-01-06 – p.22/33
U R ChatGPT ⇒ AI ( ) 2024 9-10 P
— 2025-01-06 – p.23/33
9 P P 2024 9-10 P — 2025-01-06 – p.24/33
α β P P H0 ( ) P 0.05 (P
= 0.015) (P = 0.361) 2024 9-10 P — 2025-01-06 – p.25/33
10 H0 HA k, N, ¯ ¯ x σ2 (
) MSwithin ( )MSbetween MStotal ( SStotal dftotal ) F F 2024 9-10 P — 2025-01-06 – p.26/33
(1/3) k (1) : (2) : σ2 ( ) N(µ,
σ2) µ1 = µ2 = · · · = µk N ( ) ¯ ¯ x ¯ ¯ x = k j=1 nj i=1 xji N (j i N ) 2024 9-10 P — 2025-01-06 – p.27/33
(2/3) ( )MSwithin σ2 MSwithin = SSwithin dfwithin = k
j=1 nj i=1 (xji − ¯ xj )2 N − k ( N− ) ( )MSbetween σ2 MSbetween = SSbetween dfbetween = k j=1 nj (¯ xj − ¯ ¯ x)2 k − 1 ( −1 ) ( H0 σ2 ) 2024 9-10 P — 2025-01-06 – p.28/33
(3/3) MStotal MStotal = SStotal dftotal = k j=1 nj
i=1 (xji − ¯ ¯ x)2 N − 1 ( N − 1 ) : SStotal = SSbetween + SSwithin, dftotal = dfbetween + dfwithin F F = MSbetween MSwithin F0.05 (dfbetween, dfwithin ) < F ( H0 ) 2024 9-10 P — 2025-01-06 – p.29/33
U ( p.227) 20 4 “ U.R” ( anova() )
pp.226–227 2024 9-10 P — 2025-01-06 – p.30/33
2024 9-10 P — 2025-01-06 – p.31/33
5. (1) ( ) (2) 2025 1 16 ( )
23:59 JST ( ) Waseda Moodle (Q & A ) (1)(2) Discord 2024 9-10 P — 2025-01-06 – p.32/33
2024 9-10 P — 2025-01-06 – p.33/33