@litfamily/litgrok 1.0.13 → 1.0.15

package/.grok/skills/lit-pptx/examples/04-lecture-chalk-ko.md

@@ -0,0 +1,176 @@
+---
+tonality: chalk
+title: 기초 통계 6주차 두 집단 평균 비교와 t-검정
+subtitle: 독립 표본 t-검정의 절차와 해석
+date: 2026-10-05
+department: 예시 대학교 교양 통계
+presenter: 예시 강사
+notice: 예시 데이터 — 실제 수치로 바꿔 주세요
+---
+
+---
+layout: cover-rail
+
+# 기초 통계 6주차 두 집단 평균 비교와 t-검정
+---
+
+---
+layout: agenda
+
+## 6주차 수업 순서
+
+- **질문 세우기** 귀무가설과 대립가설, 유의수준의 뜻
+- **검정 절차** 독립 표본 t-검정의 다섯 단계와 통계량 계산
+- **예제 풀이** 두 반의 시험 점수로 직접 계산해 보기
+- **해석과 함정** p값과 효과 크기를 함께 읽는 법
+---
+
+---
+layout: sidebar-note
+
+## 가설 검정의 기본 질문
+
+- **출발점: 차이가 우연으로 생길 만한 크기인가**
+  (1) 같은 시험을 다시 보면 두 반의 평균은 매번 조금씩 달라진다
+  (2) 검정은 지금의 차이가 그 흔들림보다 충분히 큰지를 묻는다
+- **귀무가설과 대립가설**
+  (1) 귀무가설: 두 모집단의 평균이 같다 (차이는 0)
+  (2) 대립가설: 평균이 다르다 (양측) 또는 한쪽이 크다 (단측)
+- **유의수준**
+  (1) 귀무가설이 맞는데도 기각할 위험을 자료를 보기 전에 정한 값
+  (2) 보통 0.05, 의학처럼 오판 비용이 큰 분야는 0.01도 쓴다
+- **판단과 보고**
+  (1) p값이 유의수준보다 작으면 귀무가설을 기각한다
+  (2) 기각하지 못한 것은 평균이 같다는 증거가 아니다
+
+::: main-box
+흔한 오해: p값은 귀무가설이 참일 확률이 아니다. 귀무가설이 참이라고 가정했을 때 지금만큼 또는 더 극단적인 차이가 나올 확률이다.
+:::
+---
+
+---
+layout: section-numeral
+
+# 검정 절차
+---
+
+---
+layout: step-diagram
+
+## 독립 표본 t-검정 다섯 단계
+
+- **1. 가설 설정** 귀무가설과 대립가설을 쓰고 양측인지 단측인지 정한다
+- **2. 가정 점검** 두 표본이 독립인지, 분포가 크게 치우치지 않았는지 본다
+- **3. 통계량 계산** 평균 차이를 표준오차로 나눠 t 값을 구한다
+- **4. p값 구하기** 자유도에 맞는 t 분포에서 t 값보다 극단적인 넓이를 찾는다
+- **5. 결론 쓰기** 기각 여부와 함께 평균 차이, 신뢰구간, 효과 크기를 적는다
+- 1-2단계는 자료를 보기 전에 끝낸다. 결과를 본 뒤 가설이나 방향을 바꾸지 않는다
+- 3-4단계는 계산기나 통계 프로그램이 해 주지만, 손으로 한 번은 따라가 본다
+- 5단계의 문장 예: "A반 평균이 4.7점 높았다 (t = 2.07, p = 0.043, d = 0.54)"
+- 출처: 예시 대학교 교양 통계 강의 노트 6장, 2026년 (예시)
+---
+
+---
+layout: method
+
+## t 통계량의 구성 요소
+
+t = (x̄₁ − x̄₂) ÷ √(s₁²/n₁ + s₂²/n₂)
+
+- **x̄₁, x̄₂** 두 표본의 평균. 예제에서는 74.2점과 69.5점
+- **x̄₁ − x̄₂** 분자, 우리가 확인하려는 평균 차이. 예제에서는 4.7점
+- **s₁², s₂²** 두 표본의 분산. 퍼짐이 클수록 분모가 커져 t 값이 작아진다
+- **n₁, n₂** 두 표본의 크기. 클수록 분모가 작아져 같은 차이도 t 값이 커진다
+- **s²/n** 각 반 평균의 흔들림. A반 8.1²/30 ≈ 2.19, B반 9.4²/30 ≈ 2.95
+- **분모** 평균 차이의 표준오차. √(2.19 + 2.95) ≈ 2.27점
+- **t** 평균 차이가 표준오차의 몇 배인지. 4.7 ÷ 2.27 ≈ 2.07
+- **자유도** 웰치 공식으로 계산, 예제에서는 약 56.6
+- **임계값** 자유도 57, 양측 0.05에서 약 ±2.00. |t|가 이보다 크면 기각
+- **웰치 검정** 두 반의 분산이 같다고 가정하지 않는 형태, 기본값으로 쓴다
+- **효과 크기 d** 평균 차이를 두 반의 합동 표준편차로 나눈 값, 예제 약 0.54
+- **판단** t ≈ 2.07이 임계값 2.00보다 커서, 두 반의 평균 차이를 우연만으로 보기 어렵다
+- 출처: 예시 대학교 교양 통계 강의 노트 6장, 2026년 (예시)
+---
+
+---
+layout: table-insight
+
+## 예제 데이터: 두 반의 기말 시험 점수
+
+| 항목 | A반 | B반 |
+|---|---|---|
+| 학생 수 | 30명 | 30명 |
+| 평균 | 74.2점 | 69.5점 |
+| 표준편차 | 8.1점 | 9.4점 |
+| 중앙값 | 75점 | 70점 |
+| 최저·최고 | 55-91점 | 48-88점 |
+| 수업 방식 | 문제 풀이 중심 | 강의 중심 |
+
+> 표 1. 두 반의 기말 시험 점수 요약, 100점 만점 (예시 데이터)
+
+- 평균 차이는 4.7점, 표준오차는 약 2.27점이다
+- t = 4.7 ÷ 2.27 ≈ 2.07, 자유도 약 56.6에서 양측 p ≈ 0.043
+- 유의수준 0.05에서 귀무가설을 기각하지만 차이의 크기는 따로 본다
+- 출처: 예시 대학교 교양 통계 수업 자료, 2026년 1학기 (예시)
+---
+
+---
+layout: comparison
+
+## 양측 검정과 단측 검정의 차이
+
+:::: columns 1fr 1fr
+::: col
+- **양측 검정**
+  (1) 질문: 두 평균이 다른가
+  (2) 기각 영역: 양쪽 꼬리에 0.025씩
+  (3) 예제 결과: p ≈ 0.043, 기각
+  (4) 쓰는 때: 방향을 미리 정할 근거가 없을 때
+:::
+::: col
+- **단측 검정**
+  (1) 질문: A반 평균이 더 큰가
+  (2) 기각 영역: 오른쪽 꼬리에 0.05
+  (3) 예제 결과: p ≈ 0.022, 기각
+  (4) 쓰는 때: 자료를 보기 전에 방향을 정했을 때
+:::
+::::
+
+- 같은 자료에서 단측 p값은 양측 p값의 절반이다
+- 그래서 방향은 자료를 보기 전에 정해 두고, 보고서에 그 근거를 적는다
+- 주: 자료를 본 뒤 단측으로 바꾸면 유의수준이 사실상 두 배가 된다
+---
+
+---
+layout: quote
+
+“p값은 연구 가설이 참일 확률을 재지 않으며, 자료가 우연만으로 생겼을 확률도 재지 않는다.”
+
+— 미국통계학회 p값 성명 원칙 2, 2016년 (번역)
+---
+
+---
+layout: image-split
+
+## t 분포와 정규 분포의 꼬리 비교
+
+![자유도 5, 30인 t 분포와 표준정규 분포 (예시) | 출처: 예시 강의 그림](assets/example-a.png)
+
+- **모양** t 분포는 정규 분포보다 꼬리가 두껍다
+- **자유도 5** 양측 0.05 임계값 2.57, 정규 분포 1.96보다 훨씬 크다
+- **자유도 30** 임계값 2.04로 정규 분포에 가까워진다
+- **뜻** 표본이 작을수록 같은 평균 차이도 유의하게 나오기 어렵다
+- **효과 크기** 예제의 코언 d는 약 0.54, 중간 정도의 차이다
+---
+
+---
+layout: closing-summary-list
+
+## 6주차 정리와 7주차 준비
+
+- t-검정은 평균 차이를 우연에 의한 흔들림의 크기와 비교하는 방법이다
+- 결론에는 p값, 평균 차이, 신뢰구간, 효과 크기를 함께 적는다
+- 방향은 자료를 보기 전에 정하고, 바꾸지 않는다
+- 과제: 교재 6장 연습문제 3·5·8번, 10월 12일 자정까지 제출 (조교 확인)
+- 7주차: 대응 표본 t-검정, 같은 학생의 사전·사후 점수 비교
+---

package/.grok/skills/lit-pptx/examples/05-data-review-night-en.md

@@ -0,0 +1,221 @@
+---
+tonality: night
+title: Mobile banking app product data review, Q3 2026
+subtitle: Activation, retention and release health for the product council
+date: 2026-10-05
+department: Example Bank, Digital Product Analytics
+presenter: Example Analytics Lead
+notice: Sample data — replace with real figures
+---
+
+---
+layout: cover-figures
+
+# Mobile banking app product data review, Q3 2026
+---
+
+---
+layout: kpi-over-chart
+
+## Q3 headline metrics and weekly active users
+
+| Weekly active users | 30-day retention | Activation rate | Crash-free sessions | App rating |
+|---|---|---|---|---|
+| 1.84 M | 61.2 % | 47.5 % | 99.62 % | 4.4 |
+| Q2 1.71 M (+7.6 %) | Q2 58.9 % | target 50 % | target 99.70 % | Q2 4.3, 2,310 reviews |
+
+::: chart type=line unit="M users"
+| Week | Weekly active users (M) |
+|---|---|
+| W27 | 1.72 |
+| W29 | 1.75 |
+| W31 | 1.77 |
+| W33 | 1.79 |
+| W35 | 1.81 |
+| W37 | 1.83 |
+| W39 | 1.84 |
+> Weekly active users, signed-in sessions, weeks 27-39 of 2026 (sample data)
+:::
+
+- Growth held at about 0.6 % a week; activation and crash-free sessions both missed target
+- Source: sample product analytics warehouse, events table, 1 July-30 September 2026 (sample)
+---
+
+---
+layout: dashboard-grid
+
+## Activation funnel, stability and session length
+
+::: chart type=bar unit="%"
+| Step | Share of installs (%) |
+|---|---|
+| Install | 100 |
+| Account linked | 72 |
+| Identity verified | 58 |
+| First transfer | 47.5 |
+> Activation funnel, Q3 installs (sample data)
+:::
+
+::: chart type=line unit="%"
+| Month | Crash-free sessions (%) |
+|---|---|
+| Jul | 99.71 |
+| Aug | 99.55 |
+| Sep | 99.60 |
+> Crash-free sessions by month (sample data)
+:::
+
+::: chart type=column unit="min"
+| Month | Median session (min) |
+|---|---|
+| Jul | 3.1 |
+| Aug | 3.3 |
+| Sep | 3.2 |
+> Median session length (sample data)
+:::
+
+- Identity verification loses 14 points of installs; the August dip follows release 8.4
+- Source: sample product analytics warehouse and crash reporter, Q3 2026 (sample)
+---
+
+---
+layout: chart-insight
+
+## 30-day retention by signup cohort
+
+::: chart type=line unit="%"
+| Signup month | Retained day 30 (%) | Retained day 60 (%) | Retained day 90 (%) |
+|---|---|---|---|
+| Mar | 56.0 | 49.1 | 45.2 |
+| Apr | 57.2 | 50.0 | 46.1 |
+| May | 58.1 | 51.4 | 47.0 |
+| Jun | 59.4 | 52.2 | 47.9 |
+| Jul | 60.3 | 53.0 | — |
+| Aug | 61.2 | — | — |
+> Share of each signup cohort still active after 30, 60 and 90 days (sample data)
+:::
+
+- Day-30 retention rose 5.2 points over six cohorts, about one point a month
+- Cohorts from the new onboarding (from June) keep the gain at day 60
+- Day-90 data for July and August cohorts is not yet complete
+- Source: sample product analytics warehouse, cohort table, March-September 2026 (sample)
+---
+
+---
+layout: section-field
+
+# Release health
+---
+
+---
+layout: full-chart
+
+## Median and p95 cold-start time by release
+
+::: chart type=line unit="s"
+| Release | Median (s) | p95 (s) |
+|---|---|---|
+| 8.1 | 1.42 | 3.10 |
+| 8.2 | 1.38 | 3.02 |
+| 8.3 | 1.35 | 2.94 |
+| 8.4 | 1.61 | 3.88 |
+| 8.5 | 1.40 | 3.05 |
+| 8.6 | 1.31 | 2.81 |
+> Cold-start time on Android and iOS combined, releases 8.1-8.6 (sample data)
+:::
+
+- Source: sample performance monitoring, field data from 2.1 M devices, Q3 2026 (sample)
+---
+
+---
+layout: method
+
+## Metric definitions and data pipeline
+
+- **Weekly active user** a signed-in user with at least one session in the ISO week
+- **30-day retention** share of a signup cohort with a session on days 28-34 after signup
+- **Activation** first completed transfer within 14 days of install
+- **Crash-free sessions** sessions without a fatal error, divided by all sessions
+- **Cold start** time from process launch to the first interactive frame, field data
+- **p95** the value 95 % of measured cold starts stay under
+- **Pipeline** client events land in the warehouse within 2 hours; the daily job runs at 04:00
+- **Exclusions** staff accounts and test devices, about 0.8 % of sessions
+- **Known gap** events from app versions below 7.9 lack the session identifier
+- Source: sample analytics data dictionary, version 2026-09 (sample)
+---
+
+---
+layout: comparison
+
+## Android and iOS release health in Q3
+
+:::: columns 1fr 1fr
+::: col
+- **Android**
+  (1) Crash-free sessions: 99.54 %, target 99.70 %
+  (2) p95 cold start: 3.4 s on release 8.6
+  (3) Share of active users: 58 %
+  (4) Main issue: memory crashes on devices under 3 GB RAM
+:::
+::: col
+- **iOS**
+  (1) Crash-free sessions: 99.73 %, target 99.70 %
+  (2) p95 cold start: 2.1 s on release 8.6
+  (3) Share of active users: 42 %
+  (4) Main issue: biometric prompt timeouts after OS update
+:::
+::::
+
+- Source: sample crash reporter and performance monitoring, Q3 2026 (sample)
+---
+
+---
+layout: big-number
+
+## Support contacts per 10,000 active users
+
+| Contacts per 10k users | Identity-check share | Median resolution |
+|---|---|---|
+| 38 | 41 % | 6.5 h |
+| Q2 44, target 35 | of all app contacts | Q2 9.0 h |
+
+- Contacts fell after the June onboarding change, mostly password resets
+- Identity checks remain the largest single reason for contact
+- Resolution time dropped once the help centre linked straight to chat
+- Source: sample customer support ticketing export, Q3 2026 (sample)
+---
+
+---
+layout: timeline
+
+## Q4 release and measurement plan
+
+| Date | Milestone | Owner |
+|---|---|---|
+| 21 Oct | Release 8.7 with low-memory image cache on Android | Mobile platform team |
+| 4 Nov | Identity check redesign to 10 % of new installs | Onboarding squad |
+| 25 Nov | Readout of the identity check experiment | Product analytics |
+| 9 Dec | Full rollout decision for the identity check redesign | Product council |
+| 16 Dec | Release 8.8 freeze before the holiday period | Release manager |
+
+- The experiment needs four weeks of new installs to detect a 3-point activation change
+- Release 8.7 is the last change to the Android crash rate measured in Q4
+- Source: sample product roadmap, version of 30 September 2026 (sample)
+---
+
+---
+layout: closing-decision-box
+
+## Q4 decisions for the product council
+
+- Decision requested: fund the identity check redesign and hold Android releases to a 3 GB device gate
+
+| Decision | Cost | Date | Owner |
+|---|---|---|---|
+| Identity check redesign experiment | 2 engineers, 6 weeks | 14 Oct | Product council |
+| Android low-memory crash gate in release checks | 1 engineer, 3 weeks | 14 Oct | Mobile platform team |
+| Retention target for Q4 cohorts at 62 % | none | 14 Oct | Product council |
+| Weekly release health digest to council members | none | 21 Oct | Product analytics |
+
+- Next step: the analytics lead circulates the experiment design by 11 October
+---

package/.grok/skills/lit-pptx/examples/06-image-led-atlas-en.md

@@ -0,0 +1,143 @@
+---
+tonality: atlas
+title: Riverside trail restoration, autumn 2026 field report
+subtitle: Phase 1 construction, visitor counts and the phase 2 request
+date: 2026-10-05
+department: Example City Parks Department
+presenter: Example Project Manager
+notice: Sample data — replace with real figures
+---
+
+---
+layout: cover-split-image
+
+# Riverside trail restoration, autumn 2026 field report
+---
+
+---
+layout: text-column
+
+## Phase 1 scope, schedule and budget
+
+- **Summary:** phase 1 finished two weeks late and 3 % under budget; weekend use is up by a third
+- **Scope:** 2.4 km of the 6.1 km trail, with 640 m of new boardwalk and 380 m of stabilised bank
+- **Schedule:** 2 June to 26 September; the July high water on the south steps cost two weeks
+- **Budget:** 4.62 M of 4.76 M spent: boardwalk 2.05, bank 1.34, paths 0.91, design 0.32
+- **Open items:** the platform railing finish and two drainage culverts, both due in November
+- Source: sample parks department project ledger and site diaries, June-September 2026 (sample)
+---
+
+---
+layout: image-full
+
+## North boardwalk after replacement
+
+![The 640 m boardwalk on recycled-composite decking, September 2026 | Source: sample photo](assets/example-a.png)
+---
+
+---
+layout: image-split
+
+## Bank stabilisation at the mill bend
+
+![Stone toe and willow stakes eight weeks after planting | Source: sample photo](assets/example-b.png)
+
+- **Problem** the bend lost about 1.2 m of bank a year to winter floods, 2019-2025
+- **Works** 380 m of stone toe, 2,600 live willow stakes, coir matting on the slope
+- **Early result** no measurable bank loss through the July high water (survey of 12 points)
+- **Planting** 84 % of willow stakes in leaf at week eight; target 70 %
+- **Next check** survey after the first winter flood, owner: river engineer
+---
+
+---
+layout: kpi-row
+
+## Visitor counts and trail condition, phase 1 section
+
+| Weekend visitors | Weekday visitors | Wheelchair passages | Path condition score | Reported hazards |
+|---|---|---|---|---|
+| 2,940 / day | 860 / day | 112 / week | 4.3 of 5 | 3 / month |
+| 2025 same weeks: 2,210 | 2025: 790 | 2025: 18 / week | 2025: 2.6 | 2025: 14 / month |
+
+- Weekend use rose by a third once the boardwalk reopened in late August
+- Wheelchair passages grew more than sixfold after the step-free route opened
+- Hazard reports fell from 14 to 3 a month, mostly loose boards and flooding
+- Source: sample automatic counters at two entrances, condition survey of 40 points, Aug-Sep 2025 and 2026 (sample)
+---
+
+---
+layout: figure-pair
+
+## South steps before and after rebuilding
+
+![South steps in May 2026, before works | Source: sample photo](assets/example-b.png)
+
+![South steps in September 2026, after works | Source: sample photo](assets/example-c.png)
+
+- Uneven timber steps replaced by 34 stone steps with a ramp beside them at 1:15
+- A handrail on both sides and a resting landing every 12 steps
+---
+
+---
+layout: photo-grid
+
+## Native planting at three sites
+
+![Meadow strip at the north entrance | Source: sample photo](assets/example-c.png)
+
+![Wet woodland edge at the mill bend | Source: sample photo](assets/example-a.png)
+
+![Shade planting under the rail bridge | Source: sample photo](assets/example-b.png)
+
+- 11,400 plants of 26 native species; 91 % survival at the September count (sample)
+- Meadow strip 4,800 plants (94 % alive), wet woodland edge 3,900 (88 %), shade planting 2,700 (90 %)
+- The wet woodland edge lost most plants where the July high water stood for four days
+---
+
+---
+layout: asymmetric-feature
+
+## Accessible viewing platform at the old mill
+
+![Viewing platform over the weir, step-free from the boardwalk | Source: sample photo](assets/example-a.png)
+
+- **Size** 6 m by 4 m, room for two wheelchairs and a bench
+- **Access** level from the boardwalk, no gap over 6 mm
+- **Use** about 310 visits a weekend day, sample counter
+- **Open item** railing finish due 14 November
+- **Cost** 0.27 M, inside the boardwalk budget
+- Note: weekend visits from the sample gate counter, September and October
+---
+
+---
+layout: timeline
+
+## Remaining construction and first-year maintenance
+
+| Date | Work | Owner |
+|---|---|---|
+| Nov 2026 | Platform railing finish and two culverts at the south steps | Contractor |
+| Dec 2026 | Post-flood bank survey at 12 points | River engineer |
+| Mar 2027 | Replacement of failed willow stakes, about 400 | Volunteer group |
+| Jun 2027 | One-year condition survey of path and boardwalk | Parks inspector |
+| Sep 2027 | Phase 1 close-out report to the council | Project manager |
+
+- Maintenance in year one costs about 46,000 from the parks revenue budget
+- Three of the five items fall to parks staff; the culverts and the willow replacement need a contractor and volunteers
+- Source: sample parks department maintenance plan, September 2026 (sample)
+---
+
+---
+layout: closing-contact-split
+
+## Phase 2 funding request, mill bridge to the weir
+
+| Item | Amount (M) | Decision by | Owner |
+|---|---|---|---|
+| Boardwalk over the wet meadow, 520 m | 1.84 | 20 Nov | Project manager |
+| Bank works at two bends, 260 m | 0.96 | 20 Nov | River engineer |
+| Path regrading and drainage, 1.4 km | 0.71 | 20 Nov | Parks inspector |
+| Design and supervision | 0.29 | 20 Nov | Project manager |
+
+- Next step: the project manager presents the phase 2 design to the council committee on 20 November
+---