Data Science Report Ϗδωεʹ͓͚Δʮग़ձ͍ʯͷ෼ੳ Vol.1 03 Tomonori Manabe R&D Group Researcher, Sansan, Inc.

01ɹ|ɹ © Sansan, Inc. Data Science Report Ϗδωεʹ͓͚Δʮग़ձ͍ʯͷ෼ੳ Vol.1 1ɹ֓ཁ ࣾձੜ׆ʹ͓͍ͯɺਓ຺ʹΑͬͯ΋ͨΒ͞ΕΔԸܙ͸গͳ͘ͳ͍ɻͱΓΘ͚Ϗδωεʹ͓͍ͯ͸ɺਓ຺ܗ੒͕Ϗδωεઓུ ͷҰͭͱ࣮ͯ͠ߦ͞Ε͍ͯΔɻྫ͑͹ɺӦۀઓུͱͯ͠λʔήοτاۀͷҙࢤܾఆऀ΍Ωʔύʔιϯͱͷਓ຺Λܗ੒͠ɺࣗࣾ αʔϏεΛೝ஌ͤ͞ɺऔҾͷܾఆΛಇֻ͖͚ΔɻӦۀ୲౰ऀ͸ɺΑΓ஍Ґͷߴ͍ҙࢥܾఆϓϩηεͷ্Ґʹ͋Δऀͱ઀৮͠Α ͏ͱ౒ྗ͠ɺͦͷΑ͏ͳΞΫηεܦ࿏Λ࡞ΔͨΊͷΞϓϩʔνΛֻ͚͍ͯΔɻ ·ͨɺاۀؒऔҾͱاۀؒͷਓ຺΋ີ઀ͳؔ܎ʹ͋Δɻ2ࣾؒͷऔҾؔ܎ʹ͸ɺ2ࣾؒͷਓ຺ͷ֦େ΍ͦͷ಺༰͕൓ө͞Ε ΔͩΖ͏ɻ2͕ࣾڠۀΛߦ͑͹ɺ্ҐΫϥεؒͷਓ຺͕֦େ͢Δ͔΋͠Εͳ͍͠ɺಉҰΫϥεʹ͓͚Δਓ຺͕ෆۉߧͰ͋Δͳ Β͹ɺ2͕ࣾݩ੥͚ͱԼ੥͚ͷؔ܎ʹ͋Δ͜ͱΛ൓ө͍ͯ͠Δͷ͔΋͠Εͳ͍ɻ ਓ຺ͱϏδωεͷؒʹ͸ɺີ઀ͳؔ܎ʹ͋Δ͜ͱ͕ظ଴͞ΕΔɻͦͷͨΊɺͦͷؔ܎ੑͷఆྔධՁ͸ɺӦۀઓ্ུॏཁͳࢦ ਑ͱͳΓಘΔɻ͔͠͠ɺͦͷΑ͏ͳఆྔԽ͸Ұൠʹ೉͍͠ͱ͞Ε͍ͯΔɻैདྷɺϏδωεਓ຺Λਖ਼֬ʹද͢Α͏ͳϏοάσʔ λ͕গͳ͔ͬͨͨΊͰ͋ΔɻSFAʢӦۀࢧԉʣπʔϧ͸ɺۙ೥ଟ͘ͷاۀͰಋೖ͕ਐΜͰ͍Δ͕ɺਓͷखʹΑΔೖྗʹΑͬͯ ͍ΔͨΊɺೖྗ࿙Ε΍༳ΕͳͲͷ໘Ͱσʔλͱͯ͠ͷਖ਼֬͞ʹෆे෼ͳ໘͕͋Δɻ·ͨɺҊ݅Խ͍ͯ͠Δ͔Ͳ͏͔ʹ͔͔ΘΒ ͣɺશͯͷࡉ͔ͳਓ຺·ͰΛ΋໢ཏతʹؚΜͰ͍ΔσʔλͰ͋Δͱ͸ݴ͍೉͘ɺਓ຺Λ෼ੳ͢ΔͨΊͷϏοάσʔλͱͯ͠͸ ෆे෼ͳ͜ͱ΋ଟ͍Ͱ͋Ζ͏ɻ Ϋϥ΢υ໊ࢗ؅ཧαʔϏεʮSansanʯͰ͸ɺϏδωε্Ͱ໘ձͨ͠ਓ෺ʹؔ͢Δਖ਼֬ͳ৘ใ͕ɺߴ͍ਫ਼౓Ͱอଘ͞ΕΔɻ ໊ࢗʹهࡌ͞Εͨݸਓ໊΍ॴଐઌʹؔ͢Δ৘ใʹՃ͑ͯɺ໊͕ࢗσʔλԽ͞Εͨ࣌ؒ৘ใ΋อଘ͞ΕΔͨΊɺ୭ʹͲͷΑ͏ͳ ॱংͰձͬͨͷ͔ʹ͍ͭͯ΋೺Ѳ͢Δ͜ͱ͕Ͱ͖Δɻਓ຺ܗ੒ͱӦۀചΓ্͛ͷؔ܎Λ෼ੳ͠ɺӦۀઓུͷݕূΛߦ͏্Ͱɺ Sansanʹอଘ͞Εͨσʔλ͸༗ӹͳσʔλϦιʔεʹͳΓಘΔͱߟ͑ΒΕΔɻ Ӧۀઓུ͸ɺݱ৔ʹ͓͍ͯܦݧʹجͮ͘͞·͟·ͳԾઆͷԼͰ࣮ߦ͞Ε͍ͯΔɻ্Ґ໾৬ऀͷΈʹूதͯ͠ΞϓϩʔνΛֻ ͚͍ͯ͘΂͖ͳͷ͔ɺͦΕͱ΋ෳ਺ͷΫϥεʹ෯޿͘઀৮͢΂͖ͳͷ͔ɺ͋Δಛఆͷ෦ॺͷΈΛ߈ΊΕ͹ྑ͍ͷ͔ɺ·ͨ͸ଟ ༷ͳ෦ॺ΁ͷΞϓϩʔν͕ඞཁͳͷ͔ɻͦ΋ͦ΋ɺ৘ใͷྲྀಈͷ଎౓͕଎͍ݱ୅ʹ͓͍ͯɺϑΣΠεɾτΡɾϑΣΠεͷ໘ձ ΍ਓ຺ܗ੒ʹରͯ͠ɺੵۃతʹίετΛ෷͏΂͖ͳͷ͔ɻ͜͏͍ͬͨ໰͍ʹର͠ɺݱ৔ͷӦۀύʔιϯ͸ܦݧͱצʹΑΔࣗ਎ ͷ౴͑Λ࣋ͪͳ͕Βɺ·ͨ͸୳Γͳ͕ΒӦۀΛߦ͍ͬͯΔ͕ɺͦͷΑ͏ͳਓ຺ܗ੒ͷϓϩηεʹର͠ɺఆྔతͳৼΓฦΓධՁ Λ༩͑Δ͜ͱ͕Ͱ͖Ε͹ɺӦۀઓཱུҊ΁ͷҰॿͱͳΓಘΔͰ͋Ζ͏ɻ ຊߘͰ͸ɺͦͷΑ͏ͳϏδϣϯͷԼɺ·ͣ͸଍ֻ͔Γͱͯ͠ʮӦۀύʔιϯͷอ༗໊ࢗͱച্ߴʯ ʮސ٬اۀͷ໊ࢗߏ੒ͱ औҾن໛ʯʹ͍ͭͯ෼ੳΛߦͬͨɻཱͯͨԾઆ͸ʮΑΓਓͱձ͍ͬͯΔӦۀύʔιϯ͸ചΓ্͕͛ߴ͍ͷ͔ʯ ʮΑΓଟ͘ͷձ ࣾͱଟ༷ʹ઀৮͍ͯ͠ΔӦۀύʔιϯ͸ചΓ্͕͛ߴ͍ͷ͔ʯ ʮऔҾߴʹӨڹ͢Δਓ຺ཁૉ͸Կ͔ʯͳͲͰ͋Δɻ͜ΕΒͷԾ આʹର͠ɺॏճؼ෼ੳΛ༻͍ͯɺ໊ࢗʹؔ͢Δσʔλͱ֤ݸਓͷചΓ্͛੒੷ͷؔ܎Λௐ΂ͨɻ

02ɹ|ɹ © Sansan, Inc. Data Science Report 2ɹ෼ੳର৅ɺ෼ੳ݁Ռ σʔλ͸ɺSansanגࣜձࣾʹ͓͚ΔӦۀ෦һผͷചΓ্ ͛σʔλͱอ༗໊ࢗ৘ใΛ༻͍ͨɻ෼ੳର৅ͱͨ͠σʔλͷ ظؒ͸ɺ2011೥ͷ3݄͔Β2017೥ͷ5݄·Ͱͷ΋ͷͰ͋Δɻ ֤ Ӧ ۀ ෦ һ ͷ ച Γ ্ ͛ ͸ɺ ໨ త ม ਺ ͱ ͠ ͨɻ · ͨɺ Sansanʹอଘ͞Εͨσʔλ͔Βɺ૯໊ࢗຕ਺ɺ୲౰اۀͷ ϢχʔΫ਺͓Αͼձࣾଟ༷ੑࢦඪͱͯ͠γϟϊϯͷΤϯτϩ ϐʔΛࢉग़͠ɺͦΕͧΕઆ໌ม਺ͱͨ͠ɻΤϯτϩϐʔ͸ɺ อ༗໊͕ࢗಛఆͷاۀʹภ͍ͬͯΔ৔߹ʹখ͍͞஋ͱͳΔࢦ ඪ஋Ͱ͋Δɻ୲౰اۀͷϢχʔΫ਺͕ಉ਺ͷ৔߹ɺ͜ͷ਺஋ ͕খ͍͞΄Ͳɺ໊ࢗຕ਺ͷ෼෍ʹภΓ͕͋Δ͜ͱʹͳΔɻ· ͨɺ֤෦һͷ৬ҐʢϚωδϟʔ͔൱͔ʣͱ୲౰اۀن໛Λ౷ ੍͢ΔͨΊʹμϛʔม਺Λಋೖͨ͠ɻ ͜ͷ݁ՌΛද1ʹࣔ͢ɻ ෼ੳ͔Β໊ࢗຕ਺ͷ܎਺͸༗ҙʹਖ਼ͱͳΓɺSansanגࣜ ձࣾʹ͓͍ͯ͸ɺಉ͡Ӧۀ೔਺Ͱ͋Ε͹อ༗͢Δ໊ࢗຕ਺ͷ ଟ͍Ӧۀύʔιϯ͕ΑΓߴ͍ചΓ্͛Λ্͛Δ͜ͱ͕ࣔ͞Ε ͨɻ·ͨɺձࣾͷϢχʔΫ਺͕ෛͰ͋Δ͜ͱ͸ɺձࣾ౰ͨΓ ͷ໊ࢗຕ਺ͱചΓ্͛ʹ૬͕ؔ͋Δ͜ͱΛ͍ࣔͯ͠Δɻͨͩ ͠ɺଟ༷ੑͷ܎਺͸༗ҙʹਖ਼ͱͳ͓ͬͯΓɺ୲౰͍ͯ͠Δձ ࣾʹόϥϯεΑ͘ਓ຺Λ͍࣋ͬͯΔΑ͏ͳอ༗໊ࢗͷߏ੒ ͕ɺചΓ্͛ͱ૬͕ؔߴ͍͜ͱ͕໌Β͔ʹ͞Εͨɻ໊ࢗΛ1 ຕ͔͍࣋ͬͯ͠ͳ͍Α͏ͳؔ܎ͷઙ͍ձࣾΛͨ͘͞Μ୲౰͠ ͍ͯͨΓɺ·ͨ͸ಛఆͷձࣾʹภͬͨਓ຺Λ͍࣋ͬͯͨΓ͢ ΔΑ͏ͳঢ়ଶΑΓ΋ɺద౓ͳ޿͞ͱઙ͞Λ࣋ͬͨਓ຺ܗ੒Λ ͍ͯ͠Δঢ়ଶͷํ͕ྑ͍ചΓ্͛ঢ়گΛ൓ө͍ͯ͠Δͱ͍ ͑Δɻ Ҏ্ͷ͜ͱ͔Βɺอ༗໊ࢗͷߏ੒͸ݸʑͷӦۀύʔιϯͷ ύϑΥʔϚϯεʹؔ͢ΔࢦඪͱͳΓಘΔ͜ͱ͕෼͔Δɻ͜ͷ ෼ੳྫ͸ɺ͋͘·Ͱ΋ಛఆاۀͷࣄྫͰ͋Γɺม਺ͱച্ߴ ͷؔ܎ੑ͸ۀछ΍اۀʹΑͬͯҟͳΔ͜ͱ͕ߟ͑ΒΕΔɻਓ ຺ͷΈͰ͸ചΓ্͛ͱͷ૬͕ؔݟ͍ͩͤͳ͍৔߹΍ɺอ༗໊ ͕ࢗগͳ͘ɺ઀৮ྺͷઙ͍৽ناۀ਺͕ചΓ্͛ʹେ͖͘Ө ڹ͢ΔΑ͏ͳέʔεͳͲɺଟ༷ͳ݁Ռ͕͋ΓಘΔͰ͋Ζ͏ɻ ͦͷΑ͏ͳاۀಛ༗ͷ෼ੳ݁Ռʹج͍ͮͯɺ໊ࢗΛӦۀύ ϑΥʔϚϯεͷࢦඪͱͯ͠ଊΒ͑ɺͦͷՄࢹԽ΍ݸਓؒͷൺ ֱΛ௨ͯ͠ʮӦۀͷޮ཰Խʯʹ໾ཱͭ஌ݟΛಘΔ͜ͱ͕Մೳ Ͱ͋Δͱߟ͑ΒΕΔɻ ͳ͓ɺ͜Ε͸͋͘·Ͱ΋Ұఆظؒ಺ͷྦྷੵ஋ಉ࢜ͷؔ܎ੑ ʹա͗ͳ͍ͨΊɺ໊ࢗߏ੒͕ചΓ্͛ͷࢦඪͱͳΓಘΔ͜ͱ Λࣔ͢ʹͱͲ·͍ͬͯΔɻ໘ձͨ͠ਓ෺ͱͷਓ຺͕ചΓ্͛ ʹରͯ͠ͲͷΑ͏ͳҼՌؔ܎Λ༗͍ͯ͠Δͷ͔Λߟ࡯͢Δʹ ͸ɺ͞ΒʹλΠϜϥάΛ༻͍ͨ෼ੳ͕ඞཁͱͳΔɻࠓޙɺͦ ͷΑ͏ͳ࣌ܥྻؔ܎ͷ෼ੳ΋ਐΊ͍ͯ͘༧ఆͰ͋Δɻ ࣍ʹɺ ސ٬اۀ୯ҐͰɺ औҾͷ੒໿ֹۚͷେ͖͞ͱ໊ࢗσʔ λͷؔ܎ੑʹ͍ͭͯॏճؼ෼ੳΛߦͬͨɻσʔλ͔Β͸ɺ໊ Dependent variable: ݸਓͷӦۀച্߹ܭ Ӧۀ೔਺ 0.23* (0.12) ૯໊ࢗຕ਺ 0.69*** (0.20) ձࣾ਺ -0.51** (0.21) ଟ༷ੑ 0.27** (0.12) ৬Ґμϛʔ 1.27*** (0.24) ୲౰μϛʔ 0.30 (0.19) Constant -0.177*** (0.062) Observations 50 R2 0.891 Adjusted R2 0.875 Residual Std. Error 0.353 (df = 43) F Statistic 58.416*** (df = 6; 43) Note: *p<0.1; **p<0.05; ***p<0.01 ද1ɿݸਓͷ໊ࢗࢦඪͱӦۀച্߹ܭֹۚͷؔ܎

03ɹ|ɹ © Sansan, Inc. Data Science Report ࢗຕ਺ɺ෦ॺͷϢχʔΫ਺ɺ໾৬ͷϢχʔΫ਺ɺ࠷ॳʹ઀৮ ͨ͠ਓ෺ͷ໾৬Ϋϥεɺ࠷ߴ໾৬ऀͷ໾৬Ϋϥεɺ໊ࢗަ׵ Λߦͬͨ೔਺ͳͲΛઆ໌ม਺ͱͯ͠༻͍ͨɻ͜ͷ݁ՌΛද2 ʹࣔ͢ɻ ໾৬ɺ෦ॺͷछྨͷଟ͞͸ڞʹऔҾֹʹ༗ҙͳਖ਼ͷ܎਺Λ ͓ࣔͯ͠Γɺ໘ձͨ͠෦ॺ΍໾৬͕ଟ͍΄ͲɺͦͷऔҾֹ͕ ૿Ճ͢Δ͜ͱ͕෼͔ͬͨɻҰํͰɺ໊ࢗຕ਺ͷ૯ܭ͸औҾֹ ʹӨڹΛࣔ͞ͳ͔ͬͨɻ ަ׵໊ͨࢗ͠ͷதʹ࠷ߴ໾৬ऀ͕ࣾ௕ɾ෦௕Ϋϥεؚ͕· Ε͍ͯΔͱ͖ʹɺͦͷऔҾֹ͸࠷΋ߴ͘ͳΓɺ࠷ߴ໾৬ऀ͕ ՝௕ ɾ ओ೚Ϋϥεͷ৔߹ʹ͸ɺऔҾֹ͸༗ҙʹ௿Լ͍ͯͨ͠ɻ ্ҐΫϥεͱͷ໘ձཤྺ͕͋Δ΄ͲɺऔҾֹ͕େ͖͘ͳΔͷ ͸ଥ౰ͳ݁ՌͰ͋Ζ͏ɻ ·ͨɺ࠷ॳͷ໘ձऀͷ໾৬͕௿͍ͱ͖΄ͲɺऔҾֹ͸ߴ͘ ͳΔ܏޲ʹ͋ͬͨɻ࠷ॳʹݱ৔ͷਓͱ໘ձ͠ɺ۩ମతͳ՝୊ ʹର͢Δ໨ઢ߹Θ͕ͤߦΘΕ͍ͯΔͱɺେ͖ͳҊ݅΁ͱൃల ͠΍͍͢ͷ͔΋͠Εͳ͍ɻͨͩ͠ɺผͷղऍͱͯ͠ɺձࣾن ໛ͷӨڹ͕ߟ͑ΒΕΔɻ͢ͳΘͪɺҰൠʹن໛͕େ͖͍ձࣾ ΄Ͳ࠷ॳʹ্ҐΫϥεͷਓ෺ͱ໘ձ͢Δ͜ͱ͕೉͍ͨ͠Ίɺ ن໛͕େ͖ͳձࣾͱͷେ͖ͳऔҾߴ͕දΕ͍ͯΔʹա͗ͳ͍ ͔Βͱ΋ߟ͑Δ͜ͱ͕Ͱ͖ΔͨΊͰ͋Δɻձࣾن໛ͷμϛʔ ม਺Λઆ໌ม਺ʹՃ͑ͯ͸͍Δ͕ɺӨڹΛे෼ʹഉআͰ͖ͯ ͍ͳ͍ͱࢥΘΕΔɻ ͜ΕΒͷ݁Ռ͸ɺ໊ࢗ৘ใ͔ΒऔҾن໛͕ਪఆͰ͖Δ͜ ͱɺ͓ΑͼऔҾن໛ʹӨڹ͢ΔҼࢠʹ͍ͭͯͷࣔࠦΛදͯ͠ ͍Δɻ͞Βʹ෦ॺͷ৘ใ΍໘ձཤྺͷύλʔϯͳͲΛ૊Έࠐ Ή͜ͱͰɺΑΓৄࡉͳ෼ੳ͕ՄೳͱͳΔ͜ͱ΋ݟࠐΉ͜ͱ͕ Ͱ͖Δɻ Dependent variable: scale(first_money) Ҋ݅ن໛A 0.812* (0.460) Ҋ݅ن໛B 3.378*** (0.617) ໊ࢗަ׵Λͨ͠೔਺ -0.463** (0.190) ໾৬छྨ਺ 0.093** (0.039) ࠷ߴ໾৬छ: ෦௕Ϋϥε -0.027 (0.064) ࠷ߴ໾৬छ: ՝௕Ϋϥε -0.248*** (0.078) ࠷ߴ໾৬छ: ओ೚Ϋϥε -0.916*** (0.166) ࠷ߴ໾৬छ: ໾৬ͳ͠ -0.341*** (0.093) ࠷ॳͷ઀৮: ෦௕Ϋϥε 0.202*** (0.071) ࠷ॳͷ઀৮: ՝௕Ϋϥε 0.245*** (0.065) ࠷ॳͷ઀৮: ओ೚Ϋϥε 0.820*** (0.120) ࠷ॳͷ઀৮: ໾৬ͳ͠ 0.256*** (0.066) ަ׵ͨ͠ਓ਺ 0.071 (0.066) ෦ॺछྨ਺ 0.485*** (0.087) ໊ࢗຕ਺ -0.275 (0.183) Constant -0.198 (0.458) Observations 3,407 R2 0.173 Adjusted R2 0.168 Residual Std. Error 0.912 (df = 3386) F Statistic 35.320*** (df = 20; 3386) Note: *p<0.1; **p<0.05; ***p<0.01 ද2ɿܖ໿ֹۚͱ໊ࢗ৘ใͷؔ܎

04ɹ|ɹ © Sansan, Inc. Data Science Report 3ɹ݁࿦ ຊߘͰ͸ɺ֤Ӧۀύʔιϯ͕͍࣋ͬͯΔ໊ࢗͷߏ੒͕ɺͦͷӦۀύʔιϯͷചΓ্͛ͱڧؔ͘࿈͍ͯ͠Δ͜ͱɺ͓Αͼɺސ ٬اۀͱͷऔҾߴͱͦͷاۀʹର͢Δ໊ࢗͷߏ੒͕ಉ༷ʹؔ܎͍ͯ͠Δ͜ͱΛࣔͨ͠ɻ͜ͷ͜ͱ͸ɺ໊ࢗߏ੒ΛݟΔ͜ͱͰച Γ্͛੒੷΍औҾֹ͕͋Δఔ౓༧ଌͰ͖Δ͜ͱΛ͓ࣔͯ͠Γɺ͞Βʹ໊ࢗߏ੒͔ΒӦۀઓུΛཱͯΔ͜ͱ͕Ͱ͖ΔՄೳੑΛࣔ ͍ͯ͠Δɻ ಛఆͷΞΧ΢ϯτΛϕʔεͱ͠ɺاۀͷҙࢥܾఆऀͱͷؔ܎ੑΛॏࢹͨ͠ΞΧ΢ϯτɾϕʔευɾϚʔέςΟϯά͕஫໨͞ Ε͍ͯΔ͕ɺͦͷϓϩηεʹఆྔ෼ੳΛಋೖ͢Δ͜ͱͰΑΓՊֶతͳӦۀ͕ՄೳʹͳΔͱߟ͍͑ͯΔɻͦͷΑ͏ͳSFAͷ৽ ల։Λ໨ࢦ͠ɺࠓޙ΋ݚڀΛਐΊ͍͖͍ͯͨɻ

2018೥5݄31೔ ൃߦ ୲౰ݚڀһ ਅು༑ଇɹTomonori Manabe Data Science Report © Sansan, Inc. ໰͍߹Θͤ Data Science Reportࣄ຿ہ ʢSansanגࣜձࣾ ٕज़ຊ෦಺ʣ 33tech@sansan.com https://jp.corp-sansan.com/ ˞ຊࢽ͸౰ࣾαʔϏεͰఆΊΔར༻ن໿ͷڐ୚ൣғ಺Ͱಗ໊Խͨ͠σʔλΛ౷ܭతʹར༻ͯ͠ ͍·͢ɻ ˞ຊࢽ͸৘ใఏڙͷ໨తͷΈͷͨΊʹఏڙ͞ΕΔ΋ͷͰ͢ɻຊࢽΛར༻͞ΕΔํ͸ɺ ͦͷ࢖༻ʹ ͍ͭͯಠࣗʹධՁ͢Δ੹೚Λෛ͏΋ͷͱ͠ɺ ໌ࣔ·ͨ͸໧ࣔΛ໰Θͣͦͷਖ਼֬ੑɺ ׬શੑɺ ༗༻ ੑ౳ͷ͍͔ͳΔอূ΋ຊࢽʹ͸൐͍·ͤΜɻ ˞ܝࡌ͞Ε͍ͯΔ৘ใ౳͸࡞੒࣌఺ͷ΋ͷͰ͢ɻ ˞ຊࢽͷҰ෦͋Δ͍͸શ෦ΛແஅͰෳ੡ɺ సࡌɺ ෳࣸ͢Δ͜ͱΛې͡·͢ɻ