R 개발자 관점에서 바라본 지역정보 수집과 분석 전략

class: title-slide, left, bottom

# R 개발자 관점에서 바라본 지역정보 수집과 분석 전략
----
## **krvote 데이터 패키지와 Shiny 대쉬보드, 그리고 인공지능(AI)**
### 이광춘 / 한국 R 사용자회
### 2022-07-07

---
class: inverse, middle
name: aisa-tidyverse-toc

# 발표 개요

----

.pull-left[

디지털 전환이 급속도로 진행됨에 따라 데이터를 수집하고 처리할 수 없을 정도로 
**.warmyellow[데이터가 폭증]**하고 있다. 무어의 법칙에 따라 컴퓨터 하드웨어와 
네트워크 분야에서 눈부신 발전을 거듭하고 있고, 
소프트웨어 분야도 오픈소스 문화가 정착됨에 따라 빅데이터 문제에 해법을 제시하고 있다.

하지만, 데이터를 통해 가치를 창출하는 **.warmyellow[데이터 과학자]**가 병목의 중심으로 부각되고 있다. 이에 `tidyverse` 패러다임이 이런 문제를 불식시킬 수 있는 은총알(Silver Bullet)로 
제시되고 있다.

공공 지역정보 데이터 수집과 분석 전략을 **.warmyellow[`tidyverse` 패러다임]**으로 살펴보고,
특히 기술적으로 데이터 패키지와 대쉬보드, 배포와 서비스 관점을 면밀히 다뤄본다.

공공 지역정보 비정형데이터에 대한 최근 딥러닝 발전도 소개한다.
]
 
.pull-right[   
.left[

1\. [tidyverse 패러다임](#aisa-tidyverse-toc)

2\. [대한민국 공공데이터](#aisa-public-data-toc)

3\. [데이터 패키지 - krvote](#aisa-krvote-toc)

4\. [대쉬보드 - flexdashboard](#aisa-dashboard-toc)

5\. [자동화, AI, 딥러닝](#aisa-ai-toc)

]
]

---
name: aisa-tidyverse-toc
# 데이터 과학

데이터 사이언스(Data Science)는 데이터를 다루는 과학이다. 따라서 자연법칙을 다루는 자연과학과 맥이 닿아있다.
자연법칙을 측정하여 데이터프레임(`DataFrame`)으로 나타내면 **변수(Variable)**, **관측점(Observation)**, **값(Value)**으로 표현된다.

.footnote[
[Google Search - "데이터 과학"](https://www.google.com/search?q=%EB%8D%B0%EC%9D%B4%ED%84%B0+%EA%B3%BC%ED%95%99)
]

---
name: tidyverse
# AI/ML/DL vs 데이터 과학

.center[
<img src="fig/tidyvese_os.png" width="77%" />
]

.footnote[
[한국통계학회 소식지 2019년 10월호 ](https://statkclee.github.io/ds-authoring/ds-stat-tidyverse.html)
]

---
name: tidyverse-definition
# `tidyverse`

.center[
<img src="fig/maso_tidyverse.png" width="77%" />
]

.pull-left[

- 기존 자료구조를 재사용
- 파이프 연산자로 간단한 함수를 조합
- 함수형 프로그래밍을 적극 사용
- 기계가 아닌 인간을 위한 설계

]

.pull-right[
 
- Reuse existing data structures.
- Compose simple functions with the pipe.
- Embrace functional programming.
- Design for humans.

]

---
name: data-science-graph
## Big Data/AI/ML/Data Science

.pull-left[
### Google Trends: 미국

![](fig/google-trends-comparison-us-1.png)
]

.pull-right[
### Google Trends: 대한민국

![](fig/google-trends-comparison-kr-1.png)
]

.footnote[
[이광춘 (2020-07-21), "기업운영혁신을 위한 데이터 과학: 기업의 활용방안", 포항산업과학연구원(RIST)](https://statkclee.github.io/ds-authoring/ds-rist.html)
]

---
name: avengers-problem
# 펭귄 어벤저스

.center[
![](fig/shiny-python.png)
]

.footnote[
[펭귄 성별예측모형: 파이썬 + `Shiny`](https://statkclee.github.io/model/tidyverse-penguin-python-shiny.html)
]

---
names: avengers-penguin
# 펭귄 데이터셋

.center[
![](fig/penguin-dataset.png)
]

.footnote[
- [펭귄 vs 붓꽃 데이터](https://statkclee.github.io/data-science/ds-iris-penguin.html)
]

---
name: avengers-architecture
# 모형 아키텍쳐

| 앙상블 | Random Forest | XGBoost |
|--------|----------------|---------|
|![](fig/penguin-arch-ensemble.png)| ![](fig/penguin-arch-rf.png) | ![](fig/penguin-arch-xgb.png) |

.footnote[
- [펭귄 성별예측모형: tidymodels](https://statkclee.github.io/model/tidyverse-parsnip-penguin.html)
]

---
name: avengers-tuning
# 모형 최적화: Hyper Parameter Tuning

.center[
![](fig/xgboost-python.png)
]

.footnote[
- [펭귄 성별예측모형: tidymodels - Hyper Parameter](https://statkclee.github.io/model/tidyverse-parsnip-penguin-hyper-parameter.html)
]

---
name: avengers-restful
# 모형 배포: RESTful API

.center[
<img src = "fig/penguin-restful-api.png" width = "100%" />
]

.footnote[
- [펭귄 성별예측모형: tidymodels - RESTful API](https://statkclee.github.io/model/tidyverse-parsnip-penguin-RESTful-api.html)
]

---
name: avengers-explainable-ai
# 설명가능한 예측모형: Explainable AI

.center[
![](fig/mdp-DALEX.png)
]

.footnote[
- [펭귄 성별예측모형: tidymodels - MDP / DALEX](https://statkclee.github.io/model/tidyverse-parsnip-penguin-DALEX.html)
]

---
name: avengers-frontend-backed
# Shiny Front-End + `tidymodels` Back-End

.center[
![](fig/penguin-shiny-predictive-model.gif)
]

.footnote[
- [펭귄 성별예측모형: `tidymodels` - Shiny](https://statkclee.github.io/model/tidyverse-parsnip-penguin-shiny.html)
]

---
name: avengers-shiny-python
# Shiny Front-End + Python Back-End

.center[
![](fig/shiny-fe-python-be.gif)
]

.footnote[
- [펭귄 성별예측모형: 파이썬 - `Shiny`](https://statkclee.github.io/model/tidyverse-penguin-python-shiny.html)
]

---
name: avengers-model
# 데이터 과학 EDA 환경: R + 파이썬

.center[
<img src= "fig/ds-avengers-model.png" width="77%" />
]

.footnote[
- [데이터 과학 개발 환경: R+파이썬](https://statkclee.github.io/data-science/ds-languages.html)
]

---
class: inverse, middle
name: aisa-public-data-toc

# 발표 개요

----

.pull-left[

하지만, 데이터를 통해 가치를 창출하는 **.warmyellow[데이터 과학자]**가 병목의 중심으로 부각되고 있다.
이에 `tidyverse` 패러다임이 이런 문제를 불식시킬 수 있는 은총알(Silver Bullet)로 
제시되고 있다.

공공 지역정보 데이터 수집과 분석 전략을 **.warmyellow[`tidyverse` 패러다임]**으로 살펴보고,
특히 기술적으로 데이터 패키지와 대쉬보드, 배포와 서비스 관점을 면밀히 다뤄본다.
]
 
.pull-right[   
.left[

1\. [`tidyverse` 패러다임](#aisa-tidyverse-toc)

2\. **[.warmyellow[대한민국 공공데이터]](#aisa-public-data-toc)**

3\. [데이터 패키지 - `krvote`](#aisa-krvote-toc)

4\. [대쉬보드 - `flexdashboard`](#aisa-dashboard-toc)

5\. [자동화, AI, 딥러닝](#aisa-ai-toc)

]
]

---
name: asia-public-data
# 공공데이터 포털

<div id="htmlwidget-12631ba716d4881a3aa0" class="reactable html-widget" style="width:auto;height:auto;"></div>
<script type="application/json" data-for="htmlwidget-12631ba716d4881a3aa0">{"x":{"tag":{"name":"Reactable","attribs":{"data":{"지자체_목록":["공공데이터포털","강원도","경기도","경상남도","경상북도","광주광역시","대구광역시","대전광역시","부산광역시","서울특별시","세종특별자치시","울산광역시","인천광역시","전라남도","전라북도","제주도","충청남도","충청북도"],"포털명":["공공데이터포털","강원 공공데이터","경기데이터드림","경남 빅데이터 허브 플랫폼","경북 공공데이터&통계","광주광역시 빅데이터 통합플랫폼","대구 D데이터허브","x","부산 공공데이터포털","서울 열린데이터광장","x","울산광역시 데이터포털","인천광역시 데이터포털","전라남도 빅데이터허브","전라북도 빅데이터허브","제주 데이터허브","충청남도 공공데이터 개방","충청북도 빅데이터 허브 플랫폼"],"x2021_7_현황_점검":["o","o","o","o","o","o","o",null,"o","o",null,"o","o","x","o","o","o","x"],"x2022_04_유지_여부_점검":["x","x","o","x","x","x","o",null,"o","o",null,"x","x","신규","o","o","o","신규"]},"columns":[{"accessor":"지자체_목록","name":"지자체_목록","type":"character"},{"accessor":"포털명","name":"포털명","type":"character","minWidth":200},{"accessor":"x2021_7_현황_점검","name":"x2021_7_현황_점검","type":"character"},{"accessor":"x2022_04_유지_여부_점검","name":"x2022_04_유지_여부_점검","type":"character"}],"defaultPageSize":10,"paginationType":"numbers","showPageInfo":true,"minRows":1,"dataKey":"6cd1c8446261ce71c8abbac04e471ac2"},"children":[]},"class":"reactR_markup"},"evals":[],"jsHooks":[]}</script>

.footnote[
- [김학래, "2022 데이터 인사이트 세미나", 중앙대학교 HIKE 연구실](https://forum.datahub.kr/t/topic/128)   
]

---
name: asia-public-data-type
# 공공데이터 자료 제공 형식

<div id="htmlwidget-148a8a653fe0099465de" class="reactable html-widget" style="width:auto;height:auto;"></div>
<script type="application/json" data-for="htmlwidget-148a8a653fe0099465de">{"x":{"tag":{"name":"Reactable","attribs":{"data":{"포털_목록":["공공데이터포털","강원 공공데이터","경기데이터드림","경남 빅데이터 허브 플랫폼","경북 공공데이터&통계","광주광역시 빅데이터 통합플랫폼","대구 D데이터허브","부산 공공데이터포털","서울 열린데이터광장","울산광역시 데이터포털","인천광역시 데이터포털","전라남도 빅데이터허브","전라북도 빅데이터허브","제주 데이터허브","충청남도 공공데이터 개방","충청북도 빅데이터 허브 플랫폼"],"서비스_유형":["LINK, LOD, REST, RSS/ATOM,YSOAP,Y다운로드,Y배송 및 기타","LINK,YSHEET,YCHART,YOPENYAPI,YMAP,YFILE","FILE,YLINK,YSHEET,YMAP,YCHART,YOPENYAPI","-","파일,YAPI,Y데이터,Y오픈API","파일,Y링크","API,YFILE","SHEET,YMAP,YAPI,YFILE,YLINK","SHEET,YOpenAPI,YFILE,YCHART,YLINK","-","FILE,YOpenAPI","FILE,YOPENAPI,YSHEET","CSV, EXCEL, GRID, CHART,YMAP,YFILE,YLINK,YAPI","FILE,YAPI,YLINK,Y영상,Y차트","-","-"]},"columns":[{"accessor":"포털_목록","name":"포털_목록","type":"character","minWidth":60},{"accessor":"서비스_유형","name":"서비스_유형","type":"character"}],"defaultPageSize":10,"paginationType":"numbers","showPageInfo":true,"minRows":1,"dataKey":"a0d9669589ab295cbfdd1be1eca7b99e"},"children":[]},"class":"reactR_markup"},"evals":[],"jsHooks":[]}</script>

.footnote[
- [김학래, "2022 데이터 인사이트 세미나", 중앙대학교 HIKE 연구실](https://forum.datahub.kr/t/topic/128)   
]

---
name: asia-public-data-type
# 공공데이터 정해진 미래(?)

.center[
![](fig/hike-data-insight.jpg)
]

.footnote[
- [김학래, "2022 데이터 인사이트 세미나", 중앙대학교 HIKE 연구실](https://forum.datahub.kr/t/topic/128)   
]

---
class: inverse, middle
name: aisa-krvote-toc

# 발표 개요

----

.pull-left[

하지만, 데이터를 통해 가치를 창출하는 **.warmyellow[데이터 과학자]**가 병목의 중심으로 부각되고 있다.
이에 `tidyverse` 패러다임이 이런 문제를 불식시킬 수 있는 은총알(Silver Bullet)로 
제시되고 있다.

1\. [`tidyverse` 패러다임](#aisa-tidyverse-toc)

2\. [대한민국 공공데이터](#aisa-public-data-toc)

3\. **[.warmyellow[데이터 패키지 - `krvote`]](#aisa-krvote-toc)**

4\. [대쉬보드 - `flexdashboard`](#aisa-dashboard-toc)

5\. [자동화, AI, 딥러닝](#aisa-ai-toc)

]
]

---
name: asia-krvote
# 패키지 / 라이브러리 제작 이유

.center[
![](fig/asia-pkg-why.jpg)
]

.footnote[
- [Desirée De Leon, Alison Hill, "A Handbook for Teaching and Learning with R and RStudio", 2019-09-11](https://rstudio4edu.github.io/rstudio4edu-book/)]

---
name: asia-krvote-process
# R 데이터 패키지

.center[
![](fig/asia-pkg-overview.jpg)
]

.footnote[
- [Desirée De Leon, Alison Hill, "A Handbook for Teaching and Learning with R and RStudio", 2019-09-11](https://rstudio4edu.github.io/rstudio4edu-book/)]

---
name: asia-krvote-nec
# 주요 선거 일정

<script type="text/javascript">
 
// jsData 
function gvisDataTimelineID59b41203625d () {
var data = new google.visualization.DataTable();
var datajson =
[
 [
"대통령",
"15대",
new Date(1997,11,18,0,0,0),
new Date(2002,11,18,0,0,0)
],
[
"대통령",
"16대",
new Date(2002,11,19,0,0,0),
new Date(2007,11,18,0,0,0)
],
[
"대통령",
"17대",
new Date(2007,11,19,0,0,0),
new Date(2012,11,19,0,0,0)
],
[
"대통령",
"18대",
new Date(2012,11,19,0,0,0),
new Date(2017,4,9,0,0,0)
],
[
"대통령",
"19대",
new Date(2017,4,10,0,0,0),
new Date(2022,4,9,0,0,0)
],
[
"대통령",
"20대",
new Date(2022,4,10,0,0,0),
new Date(2027,4,9,0,0,0)
],
[
"국회의원",
"17대",
new Date(2004,4,30,0,0,0),
new Date(2008,4,29,0,0,0)
],
[
"국회의원",
"18대",
new Date(2008,4,30,0,0,0),
new Date(2012,4,29,0,0,0)
],
[
"국회의원",
"19대",
new Date(2012,4,30,0,0,0),
new Date(2016,4,29,0,0,0)
],
[
"국회의원",
"20대",
new Date(2016,4,30,0,0,0),
new Date(2020,4,29,0,0,0)
],
[
"국회의원",
"21대",
new Date(2020,4,30,0,0,0),
new Date(2024,4,29,0,0,0)
],
[
"지방선거",
"3회",
new Date(2002,6,1,0,0,0),
new Date(2006,5,30,0,0,0)
],
[
"지방선거",
"4회",
new Date(2006,6,1,0,0,0),
new Date(2010,5,30,0,0,0)
],
[
"지방선거",
"5회",
new Date(2010,6,1,0,0,0),
new Date(2014,5,30,0,0,0)
],
[
"지방선거",
"6회",
new Date(2014,6,1,0,0,0),
new Date(2018,5,30,0,0,0)
],
[
"지방선거",
"7회",
new Date(2018,6,1,0,0,0),
new Date(2022,5,30,0,0,0)
],
[
"지방선거",
"8회",
new Date(2022,6,1,0,0,0),
new Date(2026,5,30,0,0,0)
] 
];
data.addColumn('string','type');
data.addColumn('string','subtype');
data.addColumn('datetime','start');
data.addColumn('datetime','end');
data.addRows(datajson);
return(data);
}
 
// jsDrawChart
function drawChartTimelineID59b41203625d() {
var data = gvisDataTimelineID59b41203625d();
var options = {};
options["width"] = 600;
options["height"] = 550;

var chart = new google.visualization.Timeline(
    document.getElementById('TimelineID59b41203625d')
    );
    chart.draw(data,options);

}
 
 
// jsDisplayChart
(function() {
var pkgs = window.__gvisPackages = window.__gvisPackages || [];
var callbacks = window.__gvisCallbacks = window.__gvisCallbacks || [];
var chartid = "timeline";
 
// Manually see if chartid is in pkgs (not all browsers support Array.indexOf)
var i, newPackage = true;
for (i = 0; newPackage && i < pkgs.length; i++) {
if (pkgs[i] === chartid)
newPackage = false;
}
if (newPackage)
 pkgs.push(chartid);
 
// Add the drawChart function to the global list of callbacks
callbacks.push(drawChartTimelineID59b41203625d);
})();
function displayChartTimelineID59b41203625d() {
 var pkgs = window.__gvisPackages = window.__gvisPackages || [];
 var callbacks = window.__gvisCallbacks = window.__gvisCallbacks || [];
 window.clearTimeout(window.__gvisLoad);
 // The timeout is set to 100 because otherwise the container div we are
 // targeting might not be part of the document yet
 window.__gvisLoad = setTimeout(function() {
 var pkgCount = pkgs.length;
 google.load("visualization", "1", { packages:pkgs, callback: function() {
 if (pkgCount != pkgs.length) {
 // Race condition where another setTimeout call snuck in after us; if
 // that call added a package, we must not shift its callback
 return;
}
while (callbacks.length > 0)
callbacks.shift()();
} });
}, 100);
}
 
// jsFooter
</script>
 
 
<script type="text/javascript" src="https://www.google.com/jsapi?callback=displayChartTimelineID59b41203625d"></script>
 

 
<div id="TimelineID59b41203625d" 
 style="width: 600; height: 550;">
</div>

.footnote[
- [이광춘, "krvote - 대한민국 선거 데이터 패키지", 2022-06-09](https://ai-carpentry.github.io/krvote/)]

---
name: asia-krvote-sources
# krvote 원천 데이터

.panelset[

.panel[.panel-name[중앙선관위 - 웹사이트]
.center[ <img src="fig/asia-nec-info.jpg" alt="중앙선관위 - 웹사이트" width="80%" /> 
]
]

.panel[.panel-name[중앙선관위 - 자료실]
.center[ <img src="fig/asia-general-files.jpg" alt="중앙선관위 - 자료실" width="70%" /> 
]
]

.panel[.panel-name[공공데이터포털 - API]
.center[ <img src="fig/asia-data-portal-api.jpg" alt="공공데이터포털 - API 명세" width="80%" /> 
]
]

]

---
name: asia-krvote-workflow
# krvote 데이터 전처리 작업

.center[
![](fig/asia-data-workflow.png)
]

.footnote[
- [이광춘, "krvote - 대한민국 선거 데이터 패키지", 2022-06-09](https://ai-carpentry.github.io/krvote/)]

---
name: asia-krvote-workflow-after
# krvote 데이터 패키지 제작 작업

.center[
![](fig/asia-data-workflow-pkg.png)
]

.footnote[
- [이광춘, "krvote - 대한민국 선거 데이터 패키지", 2022-06-09](https://ai-carpentry.github.io/krvote/)]

---
name: asia-krvote-inside
# krvote 살펴보기

<div id="htmlwidget-dcd3d28fb47d0068cccb" class="reactable html-widget" style="width:auto;height:auto;"></div>
<script type="application/json" data-for="htmlwidget-dcd3d28fb47d0068cccb">{"x":{"tag":{"name":"Reactable","attribs":{"data":{"데이터셋":["by_election_2021","casting_votes","code_election","code_gusigun","code_party","code_precinct","counting_votes","early_voting","education_2010","education_2014","education_2018","election_20121219","election_20170509","election_20220309","general_2012","general_2016","general_2020","local_edu_20140604","local_edu_20180613","local_edu_gg_20140604","local_edu_seoul_20140604","local_sgg_20140604","local_sgg_20180613","local_sido_20140604","local_sido_20180613","nec_sgg_code","voter_emd","voter_gusigun","voter_precinct","voter_sido","voter_station"],"데이터설명":["서울, 부산 시장 재보궐선거","선거별 구시군별 투표수","중앙선거관리위원회 선거코드","중앙선거관리위원회 구시군코드 공공데이터포털 - 중앙선거관리위원회 구시군코드 검색어: 중앙선거관리위원회 코드정보","중앙선거관리위원회 정당코드","중앙선거관리위원회 선거구 코드","구시군, 선거구별 득표수","사전투표 현황","제5회 시도 교육감","제6회 교육감선거","제7회 교육감선거","제18대 대통령선거 2012년","제19대 대통령선거 2017년","제20대 대통령선거 2022년","제19대 국회의원선거","제20대 국회의원선거","제21대 국회의원선거","","","제6회 지방선거 - 교육감 - 경기","제6회 지방선거 - 교육감 - 서울","제6회 지방선거 - 구시군의장","제7회 지선 구시군의장","제6회 지방선거 - 지도지사","제7회 지방선거 - 시도지사","시도, 구시군 마스터 코드","선거별 구시군, 읍면동별 유권자수","선거별 구시군별 유권자수","선거별 구시군, 선거구별 유권자수","선거별 시도별 유권자수","투표구별 선거인수"]},"columns":[{"accessor":"데이터셋","name":"데이터셋","type":"character"},{"accessor":"데이터설명","name":"데이터설명","type":"character"}],"filterable":true,"defaultPageSize":7,"paginationType":"numbers","showPageInfo":true,"minRows":1,"dataKey":"7387207f772a96a16212ae4659515266"},"children":[]},"class":"reactR_markup"},"evals":[],"jsHooks":[]}</script>

.footnote[
- [이광춘, "krvote - 대한민국 선거 데이터 패키지", 2022-06-09](https://ai-carpentry.github.io/krvote/)]

---
class: inverse, middle
name: aisa-dashboard-toc

# 발표 개요

----

.pull-left[

하지만, 데이터를 통해 가치를 창출하는 **.warmyellow[데이터 과학자]**가 병목의 중심으로 부각되고 있다.
이에 `tidyverse` 패러다임이 이런 문제를 불식시킬 수 있는 은총알(Silver Bullet)로 
제시되고 있다.

1\. [`tidyverse` 패러다임](#aisa-tidyverse-toc)

2\. [대한민국 공공데이터](#aisa-public-data-toc)

3\. [데이터 패키지 - `krvote`](#aisa-krvote-toc)

4\. **[.warmyellow[대쉬보드 - `flexdashboard`]](#aisa-dashboard-toc)**

5\. [자동화, AI, 딥러닝](#aisa-ai-toc)

]
]

---
name: asia-krvote-ohmynews
# 뉴스기사

.center[
<img src= "fig/asia-ohmynews.jpeg" width="90%">
]

.footnote[
- [이광춘, "지방선거 시도지사 절반, 5월 12일 결정된다 - 역대 선거 시도지사 정당별 득표율 데이터를 통해 본 6월 1일 지방선거", 2022-03-24, 오마이뉴스](http://omn.kr/1xylg)]

---
name: asia-krvote-example
class: inverse, middle, center
# krvote 유용성

[제19대 대통령 선거 결과](https://aispiration.com/president/president-election-2017.html)

---
name: asia-krvote-president
class: inverse, middle, center
# 대통령선거 대쉬보드

[제20대 대통령 선거 대쉬보드](https://aispiration.com/president/social-dashboard.html)

---
name: asia-krvote-local-election
class: inverse, middle, center
# 지방선거 대쉬보드

[제8대 경기지사 대쉬보드](https://aispiration.com/president/gg-dashboard.html)

---
class: inverse, middle
name: aisa-ai-toc

# 발표 개요

----

.pull-left[

공공 지역정보 비정형데이터에 대한 최근 딥러닝 발전도 소개한다.
]
 
.pull-right[   
.left[

1\. [tidyverse 패러다임](#aisa-tidyverse-toc)

2\. [대한민국 공공데이터](#aisa-public-data-toc)

3\. [데이터 패키지 - krvote](#aisa-krvote-toc)

4\. [대쉬보드 - flexdashboard](#aisa-dashboard-toc)

5\. **[.warmyellow[자동화, AI, 딥러닝]](#aisa-ai-toc)**

]
]

---
name: automation-tax
# 세무사 업무 변천과정

.center[

![세무 업무 변천사](fig/tax-preparation.png)

]

.footnote[
[이광춘 (2019-12-04), "자동화(Automation) - ~~로봇~~ 자동화가 빼앗는 일자리 그리고 나", 분당 서현 청소년 수련관 ](https://statkclee.github.io/ds-authoring/ds-bundang-2019.html)
]

---
name: automation-productivity-gap
# 생산성과 임금 격차

<div id="htmlwidget-89bb722d88eca329e3c2" style="width:100%;height:67%;" class="plotly html-widget"></div>
<script type="application/json" data-for="htmlwidget-89bb722d88eca329e3c2">{"x":{"data":[{"x":[1948,1949,1950,1951,1952,1953,1954,1955,1956,1957,1958,1959,1960,1961,1962,1963,1964,1965,1966,1967,1968,1969,1970,1971,1972,1973,1974,1975,1976,1977,1978,1979,1980,1981,1982,1983,1984,1985,1986,1987,1988,1989,1990,1991,1992,1993,1994,1995,1996,1997,1998,1999,2000,2001,2002,2003,2004,2005,2006,2007,2008,2009,2010,2011,2012,2013,2014,2015,2016,2017,2018],"y":[0,0.0469,0.0112,0.00499999999999999,0.00470000000000001,0.0141,0.0204,0.0232,0.073,0.0704,0.0535999999999999,0.0483,0.0532,0.0347,0.0252,0.00170000000000003,0.0161999999999999,0.0265,0.0525,0.053,0.0608,0.0346,0.0354,0.0544,0.00860000000000005,0.0600000000000001,0.0678000000000001,0.1125,0.1424,0.1323,0.1261,0.1486,0.1872,0.2314,0.2067,0.2602,0.3318,0.3747,0.4103,0.4413,0.4799,0.5019,0.5458,0.5648,0.6428,0.6505,0.6665,0.6883,0.7337,0.7585,0.7694,0.8148,0.8651,0.8811,0.9201,0.9964,1.0873,1.1556,1.1773,1.1833,1.2,1.1945,1.2723,1.2974,1.3307,1.3256,1.3383,1.3334,1.3195,1.3511,1.3728],"text":["Year: 1948 value: 0.0000 metric: gap","Year: 1949 value: 0.0469 metric: gap","Year: 1950 value: 0.0112 metric: gap","Year: 1951 value: 0.0050 metric: gap","Year: 1952 value: 0.0047 metric: gap","Year: 1953 value: 0.0141 metric: gap","Year: 1954 value: 0.0204 metric: gap","Year: 1955 value: 0.0232 metric: gap","Year: 1956 value: 0.0730 metric: gap","Year: 1957 value: 0.0704 metric: gap","Year: 1958 value: 0.0536 metric: gap","Year: 1959 value: 0.0483 metric: gap","Year: 1960 value: 0.0532 metric: gap","Year: 1961 value: 0.0347 metric: gap","Year: 1962 value: 0.0252 metric: gap","Year: 1963 value: 0.0017 metric: gap","Year: 1964 value: 0.0162 metric: gap","Year: 1965 value: 0.0265 metric: gap","Year: 1966 value: 0.0525 metric: gap","Year: 1967 value: 0.0530 metric: gap","Year: 1968 value: 0.0608 metric: gap","Year: 1969 value: 0.0346 metric: gap","Year: 1970 value: 0.0354 metric: gap","Year: 1971 value: 0.0544 metric: gap","Year: 1972 value: 0.0086 metric: gap","Year: 1973 value: 0.0600 metric: gap","Year: 1974 value: 0.0678 metric: gap","Year: 1975 value: 0.1125 metric: gap","Year: 1976 value: 0.1424 metric: gap","Year: 1977 value: 0.1323 metric: gap","Year: 1978 value: 0.1261 metric: gap","Year: 1979 value: 0.1486 metric: gap","Year: 1980 value: 0.1872 metric: gap","Year: 1981 value: 0.2314 metric: gap","Year: 1982 value: 0.2067 metric: gap","Year: 1983 value: 0.2602 metric: gap","Year: 1984 value: 0.3318 metric: gap","Year: 1985 value: 0.3747 metric: gap","Year: 1986 value: 0.4103 metric: gap","Year: 1987 value: 0.4413 metric: gap","Year: 1988 value: 0.4799 metric: gap","Year: 1989 value: 0.5019 metric: gap","Year: 1990 value: 0.5458 metric: gap","Year: 1991 value: 0.5648 metric: gap","Year: 1992 value: 0.6428 metric: gap","Year: 1993 value: 0.6505 metric: gap","Year: 1994 value: 0.6665 metric: gap","Year: 1995 value: 0.6883 metric: gap","Year: 1996 value: 0.7337 metric: gap","Year: 1997 value: 0.7585 metric: gap","Year: 1998 value: 0.7694 metric: gap","Year: 1999 value: 0.8148 metric: gap","Year: 2000 value: 0.8651 metric: gap","Year: 2001 value: 0.8811 metric: gap","Year: 2002 value: 0.9201 metric: gap","Year: 2003 value: 0.9964 metric: gap","Year: 2004 value: 1.0873 metric: gap","Year: 2005 value: 1.1556 metric: gap","Year: 2006 value: 1.1773 metric: gap","Year: 2007 value: 1.1833 metric: gap","Year: 2008 value: 1.2000 metric: gap","Year: 2009 value: 1.1945 metric: gap","Year: 2010 value: 1.2723 metric: gap","Year: 2011 value: 1.2974 metric: gap","Year: 2012 value: 1.3307 metric: gap","Year: 2013 value: 1.3256 metric: gap","Year: 2014 value: 1.3383 metric: gap","Year: 2015 value: 1.3334 metric: gap","Year: 2016 value: 1.3195 metric: gap","Year: 2017 value: 1.3511 metric: gap","Year: 2018 value: 1.3728 metric: gap"],"type":"scatter","mode":"lines+markers","line":{"width":1.88976377952756,"color":"rgba(248,118,109,1)","dash":"solid"},"hoveron":"points","name":"gap","legendgroup":"gap","showlegend":true,"xaxis":"x","yaxis":"y","hoverinfo":"text","marker":{"autocolorscale":false,"color":"rgba(248,118,109,1)","opacity":1,"size":1.13385826771654,"symbol":"circle","line":{"width":1.88976377952756,"color":"rgba(248,118,109,1)"}},"frame":null},{"x":[1948,1949,1950,1951,1952,1953,1954,1955,1956,1957,1958,1959,1960,1961,1962,1963,1964,1965,1966,1967,1968,1969,1970,1971,1972,1973,1974,1975,1976,1977,1978,1979,1980,1981,1982,1983,1984,1985,1986,1987,1988,1989,1990,1991,1992,1993,1994,1995,1996,1997,1998,1999,2000,2001,2002,2003,2004,2005,2006,2007,2008,2009,2010,2011,2012,2013,2014,2015,2016,2017,2018],"y":[0,0.0624,0.1046,0.1174,0.1502,0.2082,0.2348,0.287,0.3389,0.3708,0.3808,0.4246,0.4538,0.4784,0.5232,0.5486,0.5832,0.6227,0.647,0.6668,0.7105,0.7439,0.7681,0.8166,0.9134,0.9096,0.8705,0.8686,0.8935,0.9282,0.9566,0.9325,0.8805,0.8736,0.877,0.8849,0.8703,0.8618,0.8725,0.8467,0.8402,0.8393,0.8237,0.8202,0.832,0.8346,0.8389,0.8276,0.8287,0.8487,0.8927,0.9198,0.9296,0.956,0.9949,1.0158,1.0056,0.9973,0.9988,1.0145,1.0139,1.093,1.11,1.0847,1.065,1.084,1.0908,1.1241,1.1439,1.1467,1.1562],"text":["Year: 1948 value: 0.0000 metric: Hourly compensation","Year: 1949 value: 0.0624 metric: Hourly compensation","Year: 1950 value: 0.1046 metric: Hourly compensation","Year: 1951 value: 0.1174 metric: Hourly compensation","Year: 1952 value: 0.1502 metric: Hourly compensation","Year: 1953 value: 0.2082 metric: Hourly compensation","Year: 1954 value: 0.2348 metric: Hourly compensation","Year: 1955 value: 0.2870 metric: Hourly compensation","Year: 1956 value: 0.3389 metric: Hourly compensation","Year: 1957 value: 0.3708 metric: Hourly compensation","Year: 1958 value: 0.3808 metric: Hourly compensation","Year: 1959 value: 0.4246 metric: Hourly compensation","Year: 1960 value: 0.4538 metric: Hourly compensation","Year: 1961 value: 0.4784 metric: Hourly compensation","Year: 1962 value: 0.5232 metric: Hourly compensation","Year: 1963 value: 0.5486 metric: Hourly compensation","Year: 1964 value: 0.5832 metric: Hourly compensation","Year: 1965 value: 0.6227 metric: Hourly compensation","Year: 1966 value: 0.6470 metric: Hourly compensation","Year: 1967 value: 0.6668 metric: Hourly compensation","Year: 1968 value: 0.7105 metric: Hourly compensation","Year: 1969 value: 0.7439 metric: Hourly compensation","Year: 1970 value: 0.7681 metric: Hourly compensation","Year: 1971 value: 0.8166 metric: Hourly compensation","Year: 1972 value: 0.9134 metric: Hourly compensation","Year: 1973 value: 0.9096 metric: Hourly compensation","Year: 1974 value: 0.8705 metric: Hourly compensation","Year: 1975 value: 0.8686 metric: Hourly compensation","Year: 1976 value: 0.8935 metric: Hourly compensation","Year: 1977 value: 0.9282 metric: Hourly compensation","Year: 1978 value: 0.9566 metric: Hourly compensation","Year: 1979 value: 0.9325 metric: Hourly compensation","Year: 1980 value: 0.8805 metric: Hourly compensation","Year: 1981 value: 0.8736 metric: Hourly compensation","Year: 1982 value: 0.8770 metric: Hourly compensation","Year: 1983 value: 0.8849 metric: Hourly compensation","Year: 1984 value: 0.8703 metric: Hourly compensation","Year: 1985 value: 0.8618 metric: Hourly compensation","Year: 1986 value: 0.8725 metric: Hourly compensation","Year: 1987 value: 0.8467 metric: Hourly compensation","Year: 1988 value: 0.8402 metric: Hourly compensation","Year: 1989 value: 0.8393 metric: Hourly compensation","Year: 1990 value: 0.8237 metric: Hourly compensation","Year: 1991 value: 0.8202 metric: Hourly compensation","Year: 1992 value: 0.8320 metric: Hourly compensation","Year: 1993 value: 0.8346 metric: Hourly compensation","Year: 1994 value: 0.8389 metric: Hourly compensation","Year: 1995 value: 0.8276 metric: Hourly compensation","Year: 1996 value: 0.8287 metric: Hourly compensation","Year: 1997 value: 0.8487 metric: Hourly compensation","Year: 1998 value: 0.8927 metric: Hourly compensation","Year: 1999 value: 0.9198 metric: Hourly compensation","Year: 2000 value: 0.9296 metric: Hourly compensation","Year: 2001 value: 0.9560 metric: Hourly compensation","Year: 2002 value: 0.9949 metric: Hourly compensation","Year: 2003 value: 1.0158 metric: Hourly compensation","Year: 2004 value: 1.0056 metric: Hourly compensation","Year: 2005 value: 0.9973 metric: Hourly compensation","Year: 2006 value: 0.9988 metric: Hourly compensation","Year: 2007 value: 1.0145 metric: Hourly compensation","Year: 2008 value: 1.0139 metric: Hourly compensation","Year: 2009 value: 1.0930 metric: Hourly compensation","Year: 2010 value: 1.1100 metric: Hourly compensation","Year: 2011 value: 1.0847 metric: Hourly compensation","Year: 2012 value: 1.0650 metric: Hourly compensation","Year: 2013 value: 1.0840 metric: Hourly compensation","Year: 2014 value: 1.0908 metric: Hourly compensation","Year: 2015 value: 1.1241 metric: Hourly compensation","Year: 2016 value: 1.1439 metric: Hourly compensation","Year: 2017 value: 1.1467 metric: Hourly compensation","Year: 2018 value: 1.1562 metric: Hourly compensation"],"type":"scatter","mode":"lines+markers","line":{"width":1.88976377952756,"color":"rgba(0,186,56,1)","dash":"solid"},"hoveron":"points","name":"Hourly compensation","legendgroup":"Hourly compensation","showlegend":true,"xaxis":"x","yaxis":"y","hoverinfo":"text","marker":{"autocolorscale":false,"color":"rgba(0,186,56,1)","opacity":1,"size":1.13385826771654,"symbol":"circle","line":{"width":1.88976377952756,"color":"rgba(0,186,56,1)"}},"frame":null},{"x":[1948,1949,1950,1951,1952,1953,1954,1955,1956,1957,1958,1959,1960,1961,1962,1963,1964,1965,1966,1967,1968,1969,1970,1971,1972,1973,1974,1975,1976,1977,1978,1979,1980,1981,1982,1983,1984,1985,1986,1987,1988,1989,1990,1991,1992,1993,1994,1995,1996,1997,1998,1999,2000,2001,2002,2003,2004,2005,2006,2007,2008,2009,2010,2011,2012,2013,2014,2015,2016,2017,2018],"y":[0,0.0155,0.0934,0.1224,0.1549,0.1941,0.2144,0.2638,0.2659,0.3004,0.3272,0.3763,0.4006,0.4437,0.498,0.5503,0.5994,0.6492,0.6995,0.7198,0.7713,0.7785,0.8035,0.871,0.922,0.9696,0.9383,0.9811,1.0359,1.0605,1.0827,1.0811,1.0677,1.105,1.0837,1.1451,1.2021,1.2365,1.2828,1.288,1.3201,1.3412,1.3695,1.385,1.4748,1.4851,1.5054,1.5159,1.5624,1.6072,1.6621,1.7346,1.7947,1.8371,1.915,2.0122,2.0929,2.1529,2.1761,2.1978,2.2139,2.2875,2.3823,2.3821,2.3957,2.4096,2.4291,2.4575,2.4634,2.4978,2.529],"text":["Year: 1948 value: 0.0000 metric: Net productivity","Year: 1949 value: 0.0155 metric: Net productivity","Year: 1950 value: 0.0934 metric: Net productivity","Year: 1951 value: 0.1224 metric: Net productivity","Year: 1952 value: 0.1549 metric: Net productivity","Year: 1953 value: 0.1941 metric: Net productivity","Year: 1954 value: 0.2144 metric: Net productivity","Year: 1955 value: 0.2638 metric: Net productivity","Year: 1956 value: 0.2659 metric: Net productivity","Year: 1957 value: 0.3004 metric: Net productivity","Year: 1958 value: 0.3272 metric: Net productivity","Year: 1959 value: 0.3763 metric: Net productivity","Year: 1960 value: 0.4006 metric: Net productivity","Year: 1961 value: 0.4437 metric: Net productivity","Year: 1962 value: 0.4980 metric: Net productivity","Year: 1963 value: 0.5503 metric: Net productivity","Year: 1964 value: 0.5994 metric: Net productivity","Year: 1965 value: 0.6492 metric: Net productivity","Year: 1966 value: 0.6995 metric: Net productivity","Year: 1967 value: 0.7198 metric: Net productivity","Year: 1968 value: 0.7713 metric: Net productivity","Year: 1969 value: 0.7785 metric: Net productivity","Year: 1970 value: 0.8035 metric: Net productivity","Year: 1971 value: 0.8710 metric: Net productivity","Year: 1972 value: 0.9220 metric: Net productivity","Year: 1973 value: 0.9696 metric: Net productivity","Year: 1974 value: 0.9383 metric: Net productivity","Year: 1975 value: 0.9811 metric: Net productivity","Year: 1976 value: 1.0359 metric: Net productivity","Year: 1977 value: 1.0605 metric: Net productivity","Year: 1978 value: 1.0827 metric: Net productivity","Year: 1979 value: 1.0811 metric: Net productivity","Year: 1980 value: 1.0677 metric: Net productivity","Year: 1981 value: 1.1050 metric: Net productivity","Year: 1982 value: 1.0837 metric: Net productivity","Year: 1983 value: 1.1451 metric: Net productivity","Year: 1984 value: 1.2021 metric: Net productivity","Year: 1985 value: 1.2365 metric: Net productivity","Year: 1986 value: 1.2828 metric: Net productivity","Year: 1987 value: 1.2880 metric: Net productivity","Year: 1988 value: 1.3201 metric: Net productivity","Year: 1989 value: 1.3412 metric: Net productivity","Year: 1990 value: 1.3695 metric: Net productivity","Year: 1991 value: 1.3850 metric: Net productivity","Year: 1992 value: 1.4748 metric: Net productivity","Year: 1993 value: 1.4851 metric: Net productivity","Year: 1994 value: 1.5054 metric: Net productivity","Year: 1995 value: 1.5159 metric: Net productivity","Year: 1996 value: 1.5624 metric: Net productivity","Year: 1997 value: 1.6072 metric: Net productivity","Year: 1998 value: 1.6621 metric: Net productivity","Year: 1999 value: 1.7346 metric: Net productivity","Year: 2000 value: 1.7947 metric: Net productivity","Year: 2001 value: 1.8371 metric: Net productivity","Year: 2002 value: 1.9150 metric: Net productivity","Year: 2003 value: 2.0122 metric: Net productivity","Year: 2004 value: 2.0929 metric: Net productivity","Year: 2005 value: 2.1529 metric: Net productivity","Year: 2006 value: 2.1761 metric: Net productivity","Year: 2007 value: 2.1978 metric: Net productivity","Year: 2008 value: 2.2139 metric: Net productivity","Year: 2009 value: 2.2875 metric: Net productivity","Year: 2010 value: 2.3823 metric: Net productivity","Year: 2011 value: 2.3821 metric: Net productivity","Year: 2012 value: 2.3957 metric: Net productivity","Year: 2013 value: 2.4096 metric: Net productivity","Year: 2014 value: 2.4291 metric: Net productivity","Year: 2015 value: 2.4575 metric: Net productivity","Year: 2016 value: 2.4634 metric: Net productivity","Year: 2017 value: 2.4978 metric: Net productivity","Year: 2018 value: 2.5290 metric: Net productivity"],"type":"scatter","mode":"lines+markers","line":{"width":1.88976377952756,"color":"rgba(97,156,255,1)","dash":"solid"},"hoveron":"points","name":"Net productivity","legendgroup":"Net productivity","showlegend":true,"xaxis":"x","yaxis":"y","hoverinfo":"text","marker":{"autocolorscale":false,"color":"rgba(97,156,255,1)","opacity":1,"size":1.13385826771654,"symbol":"circle","line":{"width":1.88976377952756,"color":"rgba(97,156,255,1)"}},"frame":null}],"layout":{"margin":{"t":46.7762557077626,"r":7.30593607305936,"b":28.5844748858448,"l":34.337899543379},"plot_bgcolor":"rgba(255,255,255,1)","paper_bgcolor":"rgba(255,255,255,1)","font":{"color":"rgba(0,0,0,1)","family":"NanumGothic","size":14.6118721461187},"title":{"text":"Productivity growth and hourly compensation growth, 1948-2018","font":{"color":"rgba(0,0,0,1)","family":"NanumGothic","size":17.5342465753425},"x":0,"xref":"paper"},"xaxis":{"domain":[0,1],"automargin":true,"type":"linear","autorange":false,"range":[1944.5,2021.5],"tickmode":"array","ticktext":["1960","1980","2000","2020"],"tickvals":[1960,1980,2000,2020],"categoryorder":"array","categoryarray":["1960","1980","2000","2020"],"nticks":null,"ticks":"outside","tickcolor":"rgba(51,51,51,1)","ticklen":3.65296803652968,"tickwidth":0.66417600664176,"showticklabels":true,"tickfont":{"color":"rgba(77,77,77,1)","family":"NanumGothic","size":11.689497716895},"tickangle":-0,"showline":false,"linecolor":null,"linewidth":0,"showgrid":true,"gridcolor":"rgba(235,235,235,1)","gridwidth":0.66417600664176,"zeroline":false,"anchor":"y","title":{"text":"","font":{"color":"rgba(0,0,0,1)","family":"NanumGothic","size":14.6118721461187}},"hoverformat":".2f"},"yaxis":{"domain":[0,1],"automargin":true,"type":"linear","autorange":false,"range":[-0.12645,2.65545],"tickmode":"array","ticktext":["0%","50%","100%","150%","200%","250%"],"tickvals":[0,0.5,1,1.5,2,2.5],"categoryorder":"array","categoryarray":["0%","50%","100%","150%","200%","250%"],"nticks":null,"ticks":"outside","tickcolor":"rgba(51,51,51,1)","ticklen":3.65296803652968,"tickwidth":0.66417600664176,"showticklabels":true,"tickfont":{"color":"rgba(77,77,77,1)","family":"NanumGothic","size":11.689497716895},"tickangle":-0,"showline":false,"linecolor":null,"linewidth":0,"showgrid":true,"gridcolor":"rgba(235,235,235,1)","gridwidth":0.66417600664176,"zeroline":false,"anchor":"x","title":{"text":"","font":{"color":"rgba(0,0,0,1)","family":"NanumGothic","size":14.6118721461187}},"hoverformat":".2f"},"shapes":[{"type":"rect","fillcolor":"transparent","line":{"color":"rgba(51,51,51,1)","width":0.66417600664176,"linetype":"solid"},"yref":"paper","xref":"paper","x0":0,"x1":1,"y0":0,"y1":1}],"showlegend":true,"legend":{"bgcolor":"rgba(255,255,255,1)","bordercolor":"transparent","borderwidth":1.88976377952756,"font":{"color":"rgba(0,0,0,1)","family":"NanumGothic","size":11.689497716895},"title":{"text":"생산성과 시간당 급여","font":{"color":"rgba(0,0,0,1)","family":"NanumGothic","size":14.6118721461187}}},"hovermode":"closest","barmode":"relative"},"config":{"doubleClick":"reset","modeBarButtonsToAdd":["hoverclosest","hovercompare"],"showSendToCloud":false},"source":"A","attrs":{"44b4440450b7":{"x":{},"y":{},"colour":{},"type":"scatter"},"44b410fe35f7":{"x":{},"y":{},"colour":{}}},"cur_data":"44b4440450b7","visdat":{"44b4440450b7":["function (y) ","x"],"44b410fe35f7":["function (y) ","x"]},"highlight":{"on":"plotly_click","persistent":false,"dynamic":false,"selectize":false,"opacityDim":0.2,"selected":{"opacity":1},"debounce":0},"shinyEvents":["plotly_hover","plotly_click","plotly_selected","plotly_relayout","plotly_brushed","plotly_brushing","plotly_clickannotation","plotly_doubleclick","plotly_deselect","plotly_afterplot","plotly_sunburstclick"],"base_url":"https://plot.ly"},"evals":[],"jsHooks":[]}</script>

.footnote[
[Economic Policy Institute, "The Productivity - Pay Gap", July 2019](https://www.epi.org/productivity-pay-gap/)
]

---
name: automation-sourcing-strategy
# 소싱전략의 변화

.center[

![소싱 전략의 변화](fig/the-end-of-outsourcing.png)

]

.footnote[
자료출처: Kinetics consulting services / Automation Anywhere 
]

---
name: automation-apollo
# 아폴로 우주선

.center[

![아폴로 계획에 대한 역할 구분](fig/man-machine-role-allocation.png)

]

.footnote[

[Mary (Missy) Cummings (2014), "Man versus Machine or Man + Machine?", Duke University and MIT](https://hal.pratt.duke.edu/sites/hal.pratt.duke.edu/files/u10/IS-29-05-Expert%20Opinion[1]_0.pdf)

]

---
name: automation-role-allocation
# 인간과 기계 업무분장

.center[

]

.footnote[

[Mary (Missy) Cummings (2017), "Informing Autonomous System Design Through the Lens of Skill-, Rule-, and Knowledge-Based Behaviors", Duke University](https://hal.pratt.duke.edu/sites/hal.pratt.duke.edu/files/u10/IS-29-05-Expert%20Opinion[1]_0.pdf)

]

---
name: automation-jobs
# Human Supervisory Control

.center[

![human-supervisory-control](fig/human-supervisory-control.png)

]

.footnote[

]

---
name: automation-paradox
# 모라벡의 역설(Moravec’s paradox)

.center[

]

.footnote[
미국 카네기 멜론 대학 (CMU) 로봇 공학자 한스 모라벡(Hans Moravec)이 1970년대에 ‘it is comparatively easy to make computers exhibit adult level performance on intelligence tests or playing checkers, and difficult or impossible to give them the skills of a one-year-old when it comes to perception and mobility’라는 표현으로 컴퓨터와 인간의 능력 차이를 역설적으로 표현하였다.
]

---
name: automation-comparison
# 중국어 방 (Chinese room)

.pull-left[
![중국어 방 주장을 모사한 도해](fig/chinese-room-argument.jpg)
]

.pull-right[

]

.footnote[
[중국어 방 역설 (Chinese room argument) - 대체 누가 중국어를 이해하고 있는가?](http://ko.experiments.wikidok.net/wp-d/592f718da44f1a4153e80611/View)
]

---
name: dl-timeline
# 기계학습 분류

.center[
<img src="fig/machine-learning-classification.png" alt="기계학습 분류" width="100%" />
]

---
name: dl-timeline
# 신경망 모형 여정

.center[
<img src="fig/dl-timeline.png" alt="timeline" width="100%" />
]

.footnote[
Liangqu Long, Xiangming Zeng (2022), "Beginning Deep Learning with TensorFlow: Work with Keras, MNIST Data Sets, and Advanced Neural Networks", Apress
]

---
name: dl-applicatin
# 딥러닝 활용사례

.panelset[

.panel[.panel-name[이미지 데이터]
.center[
<img src="fig/dl-image.png" alt="이미지 데이터" width="100%" />
]
]

.panel[.panel-name[텍스트 데이터]
.center[
<img src="fig/dl-text.png" alt="텍스트 데이터" width="100%" />
]
]

.panel[.panel-name[정형 데이터]
.center[
<img src="fig/dl-csv.png" alt="정형데이터" width="100%" />
]
]

.panel[.panel-name[...]
]

]

---
name: human-performance
# 딥러닝 알고리즘 성능 - 이미지

.footnote[
[Image Classification on ImageNet](https://paperswithcode.com/sota/image-classification-on-imagenet)
]

---
name: human-performance-nlp
# 딥러닝 알고리즘 성능 - 텍스트

.footnote[
[SQuAD2.0 - The Stanford Question Answering Dataset](https://rajpurkar.github.io/SQuAD-explorer/)
]

---
name: deep-learning-vs-software
# 소프트웨어 2.0 vs. 딥러닝 모형

.center[
![](fig/software-versions.png)
]

.footnote[
Domingos, Pedro. The master algorithm: How the quest for the ultimate learning machine will remake our world. Basic Books, 2015.
]

---
name: hidden-technical-debt
# 기계학습(딥러닝) 기여분

.center[
![](fig/hidden-technical-debt.png)
]

.footnote[
Sculley, David, et al. "Hidden technical debt in machine learning systems." Advances in neural information processing systems 28 (2015).
]

---
name: deep-learning-end2end
# 딥러닝 모형 개발 범위

.center[
<img src = "fig/end2end-deep-learning.png" width = "87%">
]

.footnote[
Eli Stevens, Luca Antiga, and Thomas Viehmann (2020), "Deep Learning with PyTorch", Manning.
]

---
name: nlp-transformers
# Attention & Transformers

.center[
<img src = "fig/transformers-attention.png" width = "100%">
]

.footnote[
[Lilian Weng (June 24, 2018), "Attention? Attention!"](https://lilianweng.github.io/posts/2018-06-24-attention/)
]

---
name: nlp-bert-gpt
# Attention & Transformers &rarr; GPT, BERT

.center[
<img src = "fig/transformers-gpt-bert.png" width = "100%">
]

.footnote[
[ratsgo's NLPBOOK](https://ratsgo.github.io/nlpbook/docs/language_model/)
]

---
name: nlp-big-picture
# 인공지능 엔진 개발

.center[
<img src = "fig/big-picture.png" width = "100%">
]

---
name: asia-goodbye
class: middle, inverse

.pull-left[
# **경청해 주셔서 감사합니다.**
 
## [한국 R 사용자회](https://r2bit.com/)

]

.pull-right[
.right[

]
]