kaggleã¡ã«ã«ãªã³ã³ãã®è¡¨å½°å¼ã¤ãã³ãã«åå ãã¦ãã¾ãã
2018/5/9ï¼æ°´ï¼ã«è¡¨é¡ã®ã¤ãã³ãã«åå ãã¦ãã¾ããã
kaggleé¢é£ã®å¬ãã¯åãã¦ãªã®ã§ã¯ã¯ã¯ã¯ã
ã話ãèããªããã¿ã¤ãã³ã°ãã¦ãã®ã¾ã¾è¼ãã¦ããã®ã§ãå¤å°ééã£ã¦ããã¨ãããããããããã¾ããããããããã容赦ä¸ããã
ã¡ã«ã«ãªã³ã³ãã®è¶£æ¨
æµ·å¤ããå
¥è³è
ãæããåæé訳ãç¨æãããªã©ãã¦ã¤ãã³ãã®ããã«1000ä¸ãããã
æ¥æ¬äººã®kagglerå°ãªããªããã£ã¨æ´»æ§åãããï¼
ã¨ããæ³ãã§ããã¦ãã¼ãå
±æããããã«ã¤ãã³ããéå¬ããã
第ä¸é¨
1ä½ã®ãäºäºº
3ã¤ã®ç°ãªããã¼ã¿ã»ãã
ãã¼ã¿ã»ããã«ã¤ã4ã¤ã®ã¢ãã«
ããã¤ã使ã£ãæ¹ãå質ãä¸ãã
ã¹ã³ã¢ãé«ãããããã«ãè²ã
ãªæ½çã試ãã
Sparseã®ãã¼ã¿ã»ããã使ã£ã¦ã4ã¢ãã«ãåæã«å¦ç¿ããã
ããã¹ãã®çµåãããããã¨ã大äºã ã£ã
宣è¨åã¨å½ä»¤åã®åå¦çããã
ã¨ã¦ãè¤éãªã³ã¼ãã使ã£ãã®ã§ç解ãã¥ãããªã£ã¦ãã
åå¦çã®å®ç¾©
ã³ã¼ãã®å®è¡ã®ä»æ¹ã®éã
宣è¨åã¨å½ä»¤åã2ã¤
GitHubã«è¼ããã®ã§ãã²è¦ã¦ãã ãã
ãã¤ãã©ã¤ã³ãæåããæå¾ã¾ã§ããæã
åå¦çã®ä»æ¹
äºæå¦çãã¹ããã³ã°ãäºæã®æ½åº
ä»ã«ã¯æåã®æ½åº1,2ã°ã©ã
ãªãã¢ã³ãµã³ãã«ããã®ãï¼
æ§ã
ãªã¢ãã«ããã¼ã¿ã»ããã使ç¨ããããã¸ã§ã¯ãã«ãªã£ã
ãã¼ã¿ã®å¤æ§æ§ãã°ã©ãã§æ示
12種é¡ã®åããããªãã¼ã¿ã»ããã1ã¤ã«ããå ´å
ãããããã®ã¯4ã¢ãã«ãã¤3ã¤ã®ãã¼ã¿ã»ããã«åãããã¨
è²ã ãªä»æãã大ä¼ã«åå ããªããé²å±ãã¦ããæ¯ã«å¦ãã§ããã¨ãããã¨ããã£ã
è¦ã¤ããä»æã
sparseMLPãfeedforward neural network
ããã¤ãã®ã«ãã´ãªãããã¨ãã«ãããã¤ãã®ã¬ãã«ã«åãã¦åæãã
ãªãMLPï¼
å¦ç¿ãæ©ã
CPUã§ãã¨ã¦ããããæ©ã
æ®éã®ã¢ã¼ããã¯ãã£ã¬ãã«ã§å¯¾å¿ã§ãã
å®è¡ã«é¢ãã¦ã¯Tensorflowã使ã£ã
3ä½ã®æ¹
2ã¤ã®ã¢ãã«ã使ç¨ãNNã¨FM
NNï¼å¦ç¿ã«ã¯40åãããããã¯ã¹ãã£ç¹å¾´
FNï¼å¦ç¿ã«ã¯20å
ç¹å¾´ã«ã¤ãã¦
ã»ã¨ãã©ã®ã¢ã¤ãã ã女æ§ã®ãã®
ãªãµã¤ã¯ã«ã·ã§ããã種é¡è±å¯ã«ã¢ã¤ãã ãåãæãã¦ãã
ç¹å®ä¸å¯è½ãªè¨å·ãæé¤ã^@|{
BLEU
Nã°ã©ã ããã§ãã¯ããããã§ç¿»è¨³ã®é¡ä¼¼æ§ãè©ä¾¡ãã
åç¹å¾´ã®èª¬æ
ã¢ã¤ãã åã¯ãã©ã³ãåãå«ã¾ãªãå ´åããå¤ããããããããé£çµãã
説æã¯64ï½72æå
NNã®ã¹ãã©ã¯ãã£ãä½ããã¨ã«æéãè²»ããã
å°ããªããããµã¤ãºãããã¨ãã«ã¢ãã«ãé »ç¹ã«ã¢ãããã¼ãããå¿
è¦ããã
女æ§ã®ãã®ã®ããã«ã¢ãã«ã®å¾®èª¿æ´ãè¡ã£ã
ã©ã¼ãã³ã°ã¬ã¼ãã®å¦ç¿ã«ãæéãè²»ããã
ããããµã¤ãºã¨ãªããã£ãã¤ã¶ã«ã¤ãã¦ã®èª¬æ
æåã¯907â907â1027â1127â1424â1424ã¨ããããã«ã ãã ã大ãããã¦ãã
ãã©ã¡ã¼ã¿ãã¨ã¦ãå¤ãããã«ãéå¦ç¿ããããã¡
ã¨ããã¯ã¯10以ä¸ã«æããå¿
è¦ããã£ã
誤åãä¿®æ£ããå¿
è¦ããã£ãã
ï¼iPhone7ã®ãã¢ã«ãã¡ãããã¨æ°åã®éã«ã¹ãã¼ã¹ããããªã©ï¼
éä¸ããã£ã¨ç°¡åã«ã§ããããã«å·¥å¤«ãã
4ä½ã®æ¹
åå¦ç
å°ãªãæ°ã®ã¤ã³ãããã¿ã¤ãã§å¯¾å¿
ããã¹ããã«ãã´ãª
使ã£ãã¢ãã«ã¯DeepFM
DeepFMã¢ãã«ã®ãªãµã¼ããä»ã¾ã§ãããããã£ã¦ãããããã®ã¢ã¼ããã¯ãã£ã®é©ç¨ããã®ãè¯ãã¨åãã£ãã®ã§ä½¿ã£ã
Text Embedding Layer
ããã¹ãåãè¾¼ã¿é¨å
IDãå
¥ãã¦Embeddingãåèªãå
¥ãã¦ãã¨ã³ã³ã¼ãã«ãããã¨æç« ã®ãããªã¯ã¹ãã§ãã
ãã¼ã«ã«ããããã¯ã°ãã¼ãã«ã®å
容ã®ããã¹ããåå¾
FMã¢ãã«
ç°ãªããã£ã¼ã«ãéã®ç¸äºä½ç¨ãã¢ãã«åãã
FMãã¢ãã«ã«ããªãããTensorflowã使ããã¨ãæ¨å¥¨ããã¦ãã
æåã®2ã¤ã¯ã»ã³ãã³ã¹ãã¯ãã«ãã¢ã¤ãã ã®èª¬æã®ã»ã³ãã³ã¹ãã¯ãã«
ãããã¯ãã¤ã¬ãã«ãªå¦çãå®è¡ããããã®ãã®
DNN Layer
Pure MLPã使ç¨
ã¬ã¹ãããã¯ã¡ãã£ã¨é
ãã®ã§ãæçµçã«ã¯å¹ççã§æ£ç¢ºãªMLPã使ããã¨ã«ãã
å¦ç¿ã®ã¡ã½ãã
Lazy Nadam
ä»ã§è©¦ãããªããã£ãã¤ã¶ããããããã¦ããã®ã§ãããå©ç¨ãã
Snapshot Ensemble
ä¸æ¬¡å
ã°ã©ãã§æµãã説æ
ãã¼ã«ã«ã®æä½å¤ãæ¢ã
ã¢ã³ãµã³ãã«ã®éãæ¹
Snapshot Ensemble
æåã®ã¨ããã¯ã§ã¯é常æ°å¤ãè¡°éãã¦ãã
注ç®ãããã®ã¯æå¾ã®ã¨ããã¯ã®æ°å¤
å¹çæ§
ã¢ãã«ã§ã¯ãFastText+AvaragePoolingã使ç¨
Snapshot Ensembleã¨LR Restartsã使ç¨
ã«ã¼ãã®å¤ã§ãã¤ã³ãã£ã³ã°ã¡ã½ããã使ç¨
GitHubã«æ²è¼ããã®ã§ããã£ããåèã«ãã¦ä¸ãã
ããã¾ã§ã®Q&A
Q)
kaggleããããã¨æã£ãã¢ããã¼ã·ã§ã³ã¯ï¼
A)
1ä½ã®æ¹A
ããããæ©æ¢°å¦ç¿ã«èå³ããã£ããã³ãã¥ããã£ã«åå ããããã«ãªã£ãããkaggleã«èå³ããããããã®æã®ä»äºãæ©æ¢°å¦ç¿ã«ç¹åãããã®ã§ãã£ãã®ã§ãããã«ç¥èãæ·±ãããã¨æã£ãã®ãä¸çªã®çç±
1ä½ã®æ¹B
7å¹´ä½åã«æ©æ¢°å¦ç¿ãå§ãããAndroid Incã¨ããã³ã¼ã¹ãåãã¦ãè¬ç¾©ã§å¦ãã ãã¨ãæ´»ããã¦ã¿ããã¨æã£ããããkaggleã¯æ¬å½ã«ããã£ã¦ãã¾ãã®ã§å¦ã¶ã®ã¯ç°¡åã ã£ããããã°ããã»ã©ç¹æ°ãä¸ããã®ã§ãããã¢ããã¼ã·ã§ã³ã«ãªãã
3ä½ã®æ¹
æ©æ¢°å¦ç¿ã¯1å¹´åã«å§ããã1å¹´åãããã®ãããèå³ãæã¡ãæåã®å¤§ä¼ã«åºãã®ã¯ä¸å½ã«ããã¨ããkaggleãéãã¦æ§ã
ãªåå è
ã¨äº¤æµãããã¨ã楽ãã
4ä½ã®æ¹
2013å¹´ã«courseraã®ææ¥ãåãã¦ãkaggleãã©ããªãã¨ãããã詳細ãå¦ãã§ããã®éã«èå³ãæ·±ããããã¯ãkaggleãéãã¦å¦ãã ãã¨ã使ã£ã¦ã¹ãã«ã試ããã®ãã¢ããã¼ã·ã§ã³ã«ãªã
Q)
ãããããã¨ã¢ããã¼ã·ã§ã³ãä¸ãããã¨ãããããã©ãã¢ããã¼ã·ã§ã³ãä¿ã¤æ¹æ³ã¯ï¼
A)
1ä½ã®æ¹B
Kaggleã«æèãéä¸ãããã¨
Q)
ã©ããã£ã¦kaggleã®ã³ã³ãã«åã¤ï¼ããããããããããã¿ãããªç§å¯ãããã°
A)
1ä½ã®æ¹A
åã¤ããã«å¤§äºãªãã¨ã¯ãkaggleã®åªåè
ãä½ãããããã¤ã¡ã¼ã¸ããã誰ã§ãã§ãããã¨ããã£ã¦ãæå³ããªãã®ã§ãèªåã®ãã£ã¦ãããã¨ãæ¹åãã¦ãããã©ããªãã¨ããã£ã¦ãããã®ãã¤ã¡ã¼ã¸ãã¦çªãé²ãã®ã大äºããã¨ã¯kaggleã®ä¸ã§ã大ä¼ã§æ±ãããããã¨ãå¤ãã£ã¦ããã®ã§ãä½ãæ±ãããã¦ãããç解ãã
1ä½ã®æ¹B
ã¾ãããã¼ã ã¡ã¤ãã1ä½ã§èªåã2ä½ã ã£ãæã«ãåã¡ããã¨æã£ãã競äºå¿ãã©ãã ãæéãè²»ããããé¢ä¿ãã¦ãããç°¡åã«çãã¯ãªãããã©ããã ãã®æéãè²»ããã¦ä¸çæ¸å½ãã
3ä½ã®æ¹
æéãè²»ãããã¨ã大äºãããã²ã¨ã¤è¨ããã®ã¯ã¤ãã¸ãã¼ã·ã§ã³ãæ³åãããã¨ãç¥èãã¹ãã«ããã大äºãã©ããã£ããçµæãããã£ã¦æ¥ãã®ããæè»ã«æ³åããªããè¡åãããã¤ã¡ã¼ã¸ã«çªãé²ã
4ä½ã®æ¹
ä¸çªå¤§äºãªã®ã¯ã©ãã ãåªåãããããæéãããããããªãã¨ã§ããªããã大ä¼ã«åã¤ããã«ã¯ãããããªã¢ããã¼ãã試ããªãã¦ã¯ãããªããå¦çãè²ã
試ãå¿
è¦ãããããããä»ã®äººã¨èªåã¨ãåããã¢ããã¼ã·ã§ã³
Q)
ã©ããã£ãã½ããã¦ã§ã¢ããã¼ãã¦ã§ã¢ã使ã£ã¦ããï¼ããã¼ãã¦ã§ã¢ã®å¶éã ã£ããé£ããå±é¢ã«ç´é¢ãããã¨ã¯ããï¼
A)
4ä½ã®æ¹
ããã¤ãå¾ãæ
å ±ã使ããªããã¢ãã«ããã«ããã
3ä½ã®æ¹
kaggleã®ã³ãã¥ããã£ã§ãã使ããã¦ãããã®ãç©æ¥µçã«æ¡ç¨ãããFMã¢ãã«ãä½æããéã®ãã¼ã«ã«ã¯è¦å´ãããPythonã®ã©ã¤ãã©ãªãããã大äºãkaggleã®ã«ã¼ãã«ãããªã使ã£ããkarnelã使ã£ãã¿ã¤ãã®ã³ã³ãã¯ã¨ã¦ã好ã
1ä½ã®æ¹B
ä¸è¬çãªãã¼ã«ã使ã£ã¦ããã大äºã ã¨æãã®ã¯èªåç¬èªã®æ©æ¢°å¦ç¿ã®ã¢ã«ã´ãªãºã ã使ç¨ãããã¨ãèªåã®ã¢ã«ã´ãªãºã ãå®è¡ããã®ã¯å¿
ãããæåã«ã¤ãªãããã®ã§ã¯ãªãããã¨ã¦ãè¯ãçµé¨ã«ãªã
1ä½ã®æ¹A
ãã¼ãã¦ã§ã¢ã«é¢ãã¦è¨ãã¨èªåã®ã¯ã¼ã¯ã¹ãã¼ã·ã§ã³ããã£ãã®ã§ããã¼ãã使ããªãã¦æ¸ãã ã®ã§æ¥½ã ã£ããå¦ç¿ããã¨ãã«å¤éãããªãã¦ã¯ãããªãã¨ãã¯ããããéè¦ã
Q)
Rawãã¼ã¿ã¯å¤§äºï¼
ï¼ç¹å¾´éã®ã¨ã³ã¸ãã¢ãªã³ã°ã«ããã¦ãçãã¼ã¿ã¯ã©ãã ã大äºãï¼
A)
1ä½ã®æ¹A
ç¹å¾´éã¨ã³ã¸ãã¢ãªã³ã°ãããªãã¨ãããã£ã大ä¼ã«åã¤ãã¨ã¯é£ãããç¹å¾´ã«é¢ãã¦è¨ãã¨å
¨ã¦Rawãã¼ã¿ããéãããã®
1ä½ã®æ¹B
ãã¼ã ã¡ã¤ãã«åæããããã ãç¹å¾´éã¨ã³ã¸ãã¢ãªã³ã°ã¯æã¨ãã¦ããããã
3ä½ã®æ¹
ã¢ãã«ã¯å¿
ããããããè³¢ãããã§ã¯ãªãã®ã§ãå¦ç¿ããããã¨ã大äºã«ãªã£ã¦ãããç¹å¾´éãè¦ã¤ããã¨ããããã¯ãå¦ç¿ãããã¦ã¢ãã«ãè³¢ããã¦ãããã¨ã大äºãå人çã«ã¯ç¹å¾´ã¨ã³ã¸ãã¢ãªã³ã°ã¯ããã大äºãããã°ããã»ã©è³ªãè¯ããªã
4ä½ã®æ¹
ãã¼ã¿ãéããéã«ãããä½ã®ç¹å¾´ãåãããªããã¨ããã£ããIDçã§ä½ã使ããªããªã©ãRawãã¼ã¿ã§ããã°ç¹å¾´éãåãããã¨ã³ã¸ãã¢ãªã³ã°ãã¦ãããã®ã§å¤§äºã«ãã¦æ¬²ãã
第äºé¨
takaptæ°ãä¸å°¾æ°ãä½è¤æ°ãã¡ã«ã«ãªä¸ç°æ°
Q)
kaggleãå§ãããã£ããã¯ï¼
A)
ä¸å°¾æ°
æªçµé¨ã§MLã¨ã³ã¸ãã¢ãç®æãã¨ãªãã¨çªç ´å£ãå¿
è¦ã«ãªãããkaggleããã£ããã«ãªã
takaptæ°
å
ã
競æããã°ã©ãã³ã°ããã¦ãããæ©æ¢°å¦ç¿ã触ã£ã¦1å¹´ãããã競ãã®ã好ããªã®ã§kaggleãèªç¶ã«å§ããã競æããã°ã©ãã³ã°ã¨kaggleã¨ã®é¢é£æ§ã¯ãããªã«ãªãã¨æãã¦ãã
Q)
kaggleããã£ã¦ãã¦ããã£ããã¨
A)
ä¸å°¾æ°
å®åã§MLãé©ç¨ãããã¨ããã¨åãã¿ã¹ã¯ãå¤ããªããã¡ãkaggleã ã¨è²ã
ãªä¸çã«è§¦ãããã¨ãã§ããã®ã§é¢ç½ãã¨æã
ä½è¤æ°
kaggleãä¸ç·ã«ãã仲éãåºæ¥ã
takaptæ°
楽ãããkaggleããã£ã¦ããã¨ãã¯çãã¦ããæããããã
â»ã¡ã«ã«ãªã§ãkaggleé¨ãåºæ¥ã¦æ´»åãã¦ããããã§ãã
Q)
ç©ãä½ã£ããè¨è¨ãããããã°ã©ãã³ã°ã«ããå½±é¿ã¯ï¼
A)
takaptæ°
kaggleã®ã¨ãã¯ã±ã±ã£ã¨åºæ¥ãããã«ãªã£ã
ä¸ç°æ°
ä¸ç1ä½ã®æ¹ã¨ã®å·®ãåããã®ãé¢ç½ã
Q)
ã¡ã«ã«ãªã®å®ãã¼ã¿ã«è¿ããã®ãã³ã³ãã§ä½¿ç¨ããããç¹å¾´ã¨ã³ã¸ãã¢ãªã³ã°ã¯ã©ãããï¼
A)
takaptæ°
èªç¶è¨èªã¨ã³ã¸ãã¢ãªã³ã°ã«é¢ãããã¨ãçã£ç«¯ãããã£ããkarnelã®æ¢åã©ã¤ãã©ãªã®ä»æ§ã®è¦è½ã¨ãã«æ°ã¥ããã¨ã大äºãæ°ã¥ããã«ãã¼ã¿æ¬ æãã¦ããããããNLPã«é¢ãã¦ã¯ã100å以ä¸ã®ç¹å¾´éã試ããããã¯ã¿ã©ã¤ã¶ä½¿ç¨
ä½è¤æ°
èªç¶è¨èªå¦çã¯åãããã£ããä¾¡æ ¼ã«é¢ãããã¬ã¹ãããã£ããä»åã¯æéå¶éããã£ãã®ã§ç¹å¾´éã¯å°ãªããããã¨ã大äºã ã¨æãã
ä¸å°¾æ°
ã³ã³ãã®karnelãåèã«ãã¦ããããããããããã®ãè¦ã¤ããã
Q)
ã¢ãã«ã®é¸å®æ¹æ³ã®åºæºã¯ï¼
A)
takaptæ°
å§ããæç¹ã§ä¸çªå¼·ãkarnelãåèã«ããããã®ãã¨ãã¢ã³ãµã³ãã«ãããè²ã
試ãããåºæ¬çã«ã¹ã³ã¢ãä¸ãã£ããã®ããªãã¼ã·ã§ã³ã¯ãã¼ã«ã«ã§ãã£ã¦ãæéå¶éé¢é£ã¯karnelã§å®æ½ããªãã¸å帰ã®æã¯ä¸¦åå¦çããã
ä½è¤æ°
ããããããã¦ã¿ã¦ãããã®ã使ã
ä¸å°¾æ°
æ¢åkarnelãåèã«ãã¦ãããä»åã§ããã¨XGBoostãã©ã¤ãGPUã®ãããªãã®ã使ã£ãããã®ãã¨ã¢ã³ãµã³ãã«
Q)
karnelãã¼ã¹ã¨ããªãã¨ã©ã£ã¡ãããï¼
A)
takaptæ°
karnelãã¼ã¹ã®æ¹ãããããªãã ã¨ã¯ã¬ã¤ã¸ã¼ã¢ã³ãµã³ãã«ãªã©ä½ã§ããããããªã®ã§
Q)
kagglerãªãã§ã¯ã®æ©ã¿ã¯ãããï¼
A)
takaptæ°
ä»äºãçµãã£ããkaggleãåæ¥ã¯kaggleãã¤ããã®ã楽ãã¿ã®ãã¡
ä½è¤æ°
ã³ã³ãã«ãã£ã¦ã¯ãã¼ã¿ã®ãµã¤ãºãã¾ã¡ã¾ã¡ãªã®ã§å¤§ãããµã¤ãºã ã¨ãªããªãåå ãã¥ããã®ããããå°ãããµã¤ãºã®ãã®ãæè¿ã¯å°ãªãã¦æ®å¿µ
ä¸å°¾æ°
ä»äºã¨kaggleä¸è²ã«ãªããï¼ãã¾ã¯ä»äºä¸ã«20ï¼
ãããæéãå²ãããâ¦ï¼
Q)
ä¸ä½å
¥è³ããã«ã¯ï¼
A)
takaptæ°
karnelãèªããªã©ããªããæéãå²ããã¨ãåãã¦åå ããQuoraã³ã³ãã§ã¯karnelã§èª¿ã¹ã¦åãããªãã£ããä»ã§ãè«æã¯ç½ ãããç´æ¥å½¹ã«ç«ã¤ãã¨ã¯ãã¾ããªãkaggleã®é ä½ã«ãç´çµããªããããç¥èã«ã¯ãªã
ä½è¤æ°
æåã«ã²ãããkarnelãè¦ãã30åãããè¦ãã¨æ¹éãç«ã¤ãããã¦ããªããããªäºã«ãçæãã
â»ãã©ãããã°ä¸ä½ã«å ¥è³ã§ãããã®å¾åã¯æãã®ã»ãä¼¼ã¦ãããããï¼
Q)
ã¡ã«ã«ãªã«è³ªåãã³ã³ãã主å¬ããå¬ããã¯ï¼
A)
å®éã«ã¡ã«ã«ãªUSã§ä½¿ã£ã¦ãããã®ãä¸ä½ã®æ¹ã®ã¯ãã¼ã¹ã©ã¤ã³ããããã¢ãã«ã§æ®éã«å¿ç¨ã§ããã¬ãã«ãåç´ã«ããã°ã©ã ã使ããã®ããããã¨
Q)
NLPã¨ãã¦ã¯åºæ¬çãªãã¨ããã¦ããããã«è¦ããããä»ã«ã¯ä½ãç¹å¥ãªãã¨ãï¼
A)
takaptæ°
ã¹ããã³ã°ãªã©ã¯ããããå
¨ãã¹ã³ã¢ãä¸ãããªãã£ãã®ã§æçµçã«ã¯ããªãã£ã
ä½è¤æ°
çµµæåãé©åãªåèªã«ç½®ãæãããã¯æéãããããkaggleã¯ä¸æ¯ãªæ¦ãã®æããã£ã¦ãã¡ãã£ã¨ããã¹ã³ã¢ã¢ãããããããã®æéãããããã¨ããã
Q)
åã³ã³ãã§å
±éã®ã¢ããã¼ãããããã¨ä½ããã£ã¦æ¹éãç«ã¦ã¦ããã®ãï¼ï¼æ¹éãç«ã¦ãã¾ã§ã®ããã»ã¹ãç¥ãããï¼
A)
takaptæ°
åé¡ãã¡ããã¨ç解ãããã¤ã³ã¼ã«ããã¼ã¿ãè¦ããä¾¡æ ¼ã®åå¸ãã©ããªã£ã¦ããããªã©ãããããkarnelãåèã«ããªããé²ãã
ä½è¤æ°
æ©æ¢°å¦ç¿ã®ã¯ã¼ã¯ããã¼ãããã¨æã£ã¦ãã¦ããã£ããåé¡ãå®ç¾©ããå¿
è¦ãããããããããã¼ã¿ããªãå ´åãããã®ã§è¨è¨ããèãããã¨ãããã¼ã¿ããã¡ãã¨ç解ãã¦ãããã¨ã大äºããã¨ã¯takaptæ°ã¨åã
Q)
æ¬æ°ã«åã¡ã«è¡ãããã¨ããæ®ãæ¥æ°ã©ããããããåå ãããªã©æ±ºãã¦ãããï¼
A)
takaptæ°
çµäº2ã«æåãããããããã°å¤§ä½ã§ããã1ãæã ã¨è¶³ããªã
ä½è¤æ°
ã ãããçµäº1ãæåããåå ãã¦ãã
Q)
ãã¼ã ã§åå ããã¨ãã®æãã¯ï¼ãã³ã³ããªãã¥ã¼ããã¦ãã人ã¨ããã§ãªã人ã¸ã®å¯¾çã¯ï¼
A)
ä½è¤æ°
å¦æ ¡ã®çå¾ã§ãã£ã¦ããã®ã§ãã¿ã¹ã¯ãæ確ã«ãããTrelloã使ã£ãã3ã¤ããããã¼ã ããã£ã¦1ãã¼ã ã¯1人ã ãããåå ãã¦ããªããã¨ããã£ã
takaptæ°
Quoraã®ã¨ãã¯ç¤¾å
ã§ä»²ã®è¯ã人ã¨ä¸ç·ã«ãã£ãã2人ãã¼ã ã ã£ãã®ã§ã´ã¼ã¹ãåé¡ã¯ãªãã£ãããã·ã«ãã¼ãåããã®ã¯ãã¼ã ã¡ã¤ãã®ããããå°éåéãéã£ãã®ããã
ããã¾ã§é§ã足ã§ãéããã¾ããã
主å¬ï¼ç»å£è
ã®çæ§ããããã¨ããããã¾ããï¼
ãããã話ãèãã¨ãkaggleãããããªãã¾ãããã