Luận văn thạc sĩ Hệ thống thông tin quản lý: Dùng học máy xếp hạng cổ phiếu bằng các chỉ số tài chính trong quá khứ

*LӟL7KLӋX 2.1 Ĉӝng lӵc và bӕi cҧnh: - ;ӃSKҥQJFәSKLӃX 6FRULQJ5DWLQJ OjSKѭѫQJSKiSFKҩPÿLӇPFәSKLӃXGӵDWUrQFiFFKӍVӕWjLFKtQKFӫD FәSKLӃX+LӋQWҥLWURQJWәFKӭFFӫDW{LÿDQJVӱGөQJSKѭѫQJSKiSFKX\rQJLDÿӇ[iF

ĈҤI HӐC QUӔC GIA TP HӖ CHÍ MINH TRѬӠNG ĈҤI HӐC BÁCH KHOA -

TRҪN HOÀNG TUҨN ANH

DÙNG HӐC MÁY XӂP HҤNG CӘ PHIӂU BҴNG CÁC CHӌ SӔ TÀI CHÍNH

&Ð1*75Î1+ĈѬӦC HOÀN THÀNH TҤI 75ѬӠ1*ĈҤI HӐC BÁCH KHOA ±Ĉ+4*-HCM

ĈҤI HӐC QUӔC GIA TP.HCM

I 7Ç1Ĉӄ TÀI:

Dùng Hӑc Máy XӃp Hҥng Cә PhiӃu Bҵng Các ChӍ Sӕ Tài Chính Trong Quá Khӭ

II NHIӊM VӨ VÀ NӜI DUNG:

KӃt hӧp SKѭѫQJSKiS +ӑFPi\ Yj3KkQWtFKWjLFKtQKÿӇ xӃp hҥng cә phiӃu bҵng các chí sӕ tài chính trong quá khӭ

III NGÀY GIAO NHIӊM VӨ: 01/02/2019

IV NGÀY HOÀN THÀNH NHIӊM VӨ: 02/06/2019

- /mQKÿҥRF{QJW\FKӭQJNKRiQ91'LUHFW và các phòng ban trong công ty ÿm OX{Q TXDQ WkP ÿӝQJ YLrQ Yj WҥR ÿLӅX NLӋQ FKR W{L WURQJ TXi WUuQKQJKLrQFӭX

- %rQFҥQK ÿyVӵJL~SÿӥFӫDJLDÿuQKEҥQEqYjQJѭӡLWKkQÿmOX{QӫQJKӝYjWҥRÿLӅXNLӋQWӕWQKҩWÿӇtôi FyWKӇWұSWUXQJQJKLrQFӭXYjKRjQWKjQKÿӅWjLQj\

'R YӅ PһW NLӃQ WKӭF Yj WKӡL JLDQ FzQ KҥQ FKӃ OXұQ YăQ FzQ QKLӅX NKLӃPNKX\ӃW7{LPRQJÿѭӧFVӵÿyQJJyS êNLӃQFӫDFiFWKҫ\F{YjPӑLQJѭӡLÿӇOXұQYăQKRjQWKLӋQKѫQ

1 7yPWҳW

ViӋc kӃt hӧp giӳa Công nghӋ thông tin và Phân tích tài chính giúp giҧm thӡi JLDQSKkQWtFKWăQJÿӝ chính xác và giҧm sӵ phө thuӝFKRjQWRjQYjRQJѭӡi phân tích Sӵ kӃt hӧSQj\ÿmÿѭӧc ӭng dөng rҩt nhiӅu trên thӃ giӟi và tҥo ra FѫKӝi cho các nhà quҧn lý quӻ quҧn lý các tài sҧn hàng tӹ Ĉ{OD

Câu hӓi nghiên cӭu tәng thӇ trong luұQYăQQj\OjOLӋu viӋc áp dөng các thuұt toán hӑc máy có thӇ ÿyQJJySJLiWUӏ trong viӋc phân biӋt các cә phiӃu hoҥt ÿӝng hiӋu quҧ ÿӇ xây dӵng danh mөFÿҫXWѭYѭӧt trӝLKѫQ91,QGH[KRһc mӝt chӍ sӕ ÿҫXWѭNKiFQKѭ91+1;LQGH[ĈӇ trҧ lӡi câu hӓi này, các câu hӓi FRQVDXÿk\VӁ ÿѭӧc nghiên cӭu:

+ KҧQăQJGӵÿRiQFӫDFiFWKXұWWRiQ KӑF Pi\ NKLSKkQELӋWJLӳDFiFFәSKLӃXKRҥWÿӝQJNpPKLӋXTXҧYjKLӋXTXҧ"

+ &yWKӇVӱGөQJFiFFKLӃQOѭӧFÿҫXWѭEҵQJFiFKVӱGөQJFiFWKXұWWRiQKӑFPi\ÿӇ[k\GӵQJGDQKPөFÿҫXWѭYѭӧWWUӝLVRYӟLWKӏWUѭӡQJKRһFPӝWFKӍVӕÿҫXWѭ"

+ &iFWKXұWWRiQKӑFPi\FyWKӇÿyQJJySYjRYLӋFOӵDFKӑQFәSKLӃXWӕWKѫQVRYӟLPӝWF{QJFөFKӑQFәSKLӃXQJүXQKLrQ"

+ &yEҵQJFKӭQJÿiQJNӇQjRYӅPӕLOLrQKӋJLӳDNKҧQăQJGӵÿRiQFӫDFiFWKXұWWRiQKӑFPi\YjOӧLQKXұQGDQKPөFÿҫXWѭÿѭӧFWҥRWӯYLӋFGӵEiRQj\"

Trong thӃ giӟi cӫa hӑc máy, có rҩt nhiӅu thuұt toán hӑc máy có thӇ ÿѭӧc sӱ dөQJÿӇ trҧ lӡi vҩQÿӅ trong luұQYăQQj\0ӝt lӵa chӑn trong sӕ FK~QJÿѭӧc nghiên cӭu trong luұQ YăQ Qj\ Naive Bayes, Logistic Regrssion, Random Forest, Support Vector Machine (SVM)

KӃt quҧ nghiên cӭu cӫa luұQYăQÿmFKRWKҩy giҧi thuұt Randomforest cho kӃt quҧ tӕW KѫQ FiF JLҧi thuұt khác khi áp dөng vӟi bӝ dӵ liӋu cӫa thӏ WUѭӡng chӭng khoán ViӋt Nam tӯ QăPÿӃn 2018 Mһc dù khҧ QăQJGӵ báo chӍ ÿҥt mӭFWUXQJEuQKQKѭQJNӃt quҧ tӯ giҧi thuұt hӑFPi\ÿmJL~SWҥo thành các danh mөc có hiӋu quҧ KѫQFKӍ sӕ tham chiӃu VNindex và VN30

ViӋc xây dӵng và ӭng dөQJÿѭӧc hӋ thӕng kӃt hӧp trên sӁ OjFăQFӭ ÿӇ phát triӇn các phҫn tiӃp theo: Mô hình xӃp hҥng theo ngành, mô hình xӃp hҥng rӫi

ro và dӵ báo

Abstracts

The combination of Information Technology and Financial Analysis reduces analysis time, increases accuracy and reduces dependence entirely on analysts This combination has been applied a lot in the world and created opportunities for fund managers to manage billions of dollars of assets

The overall research question in this dissertation is whether the application of machine learning algorithms could make a valuable contribution in distinguishing efficient stocks to build a portfolio that outperforms VNIndex

or another investment indicators such as VN30, HNXindex To answer this question, the following sub-questions will be studied:

+ Predictability of machine learning algorithms when distinguishing between inefficient and efficient stocks?

+ Could we use investment strategies by using machine learning algorithms to build an outstanding portfolio against the market or an investment index?

+ Could machine learning algorithms contribute to stock selection better than a random stock picking tool?

+ Is there any significant evidence of the link between predictability of machine learning algorithms and portfolio returns generated from this prediction?

In the world of machine learning, there are many machine learning algorithms that can be used to answer the problem in this thesis One of them is studied in this thesis: Naive Bayes, Logistic Regrssion, Random Forest, Support Vector Machine (SVM)

The results of the thesis show that the Randomforest algorithm gives better results than other algorithms when applied to the data set of Vietnam stock market from 2010 to 2018 Although the forecasting ability is only achieved average, but the results from machine learning algorithms have helped to create more effective categories of VNindex and VN30 reference index

The construction and application of the above-mentioned combined system will be the basis for developing the following sections: Sector ranking model, risk ranking model and forecast

LӡLFDPÿRDQFӫa tác giҧ LuұQYăQ

1JѭӡLFDPÿRDQ

MӨC LӨC

1 Tóm tҳt 5

2 Giӟi ThiӋu 9

3 3KѭѫQJ3KiS;Ӄp Hҥng Cә PhiӃX7URQJĈҫX7ѭ&Kӭng Khoán: 13

4 Hӑc Máy 22

5 Dùng Hӑc Máy XӃp Hҥng Cә PhiӃu: 44

6 Thí NghiӋP9jĈiQK*Li.Ӄt Quҧ: 54

7 KӃt Luұn 66

8 Danh Mөc Các Tài LiӋu Tham Khҧo: 68

2 *LӟL7KLӋX

2.1 Ĉӝng lӵc và bӕi cҧnh:

- ;ӃSKҥQJFәSKLӃX6FRULQJ5DWLQJ OjSKѭѫQJSKiSFKҩPÿLӇPFәSKLӃXGӵDWUrQFiFFKӍVӕWjLFKtQKFӫD FәSKLӃX+LӋQWҥLWURQJWәFKӭFFӫDW{LÿDQJVӱGөQJSKѭѫQJSKiSFKX\rQJLDÿӇ[iFÿӏQKWUӑQJVӕFKRFiFFKӍVӕWjL FKtQK WURQJ P{ KuQK 3KѭѫQJ SKiS FKX\rQ JLD Fy PӝW Vӕ ÿLӇP \ӃXchính:

+ &KѭD[iFÿӏQKÿѭӧF PӭFÿӝKLӋXTXҧFӫD P{KuQKÿӝFKtQK [iFFӫDm{KuQKÿӝFKtQK[iFFKRWӯQJKҥQJÿѭӧFÿiQKJLi« 

+ 3KөWKXӝFKRjQWRjQYjRFKX\rQJLDNLӃQWKӭFWUҧLQJKLӋPFҧP[~F WURQJYLӋF[iFÿӏQKWUӑQJVӕ

- 9LӋF[iFÿӏQK[ӃSKҥQJKӧSOêFӫDFәSKLӃXVӁJL~SQKjÿҫXWѭOӵDFKӑQÿѭӧFFiFFәSKLӃXWӕWYjJLDWăQJKLӋXTXҧÿҫXWѭ

+ 'DYLG+DUGLQJ{QJOjPFKRTXӻ:LQWRQ&DSLWDOWӹĈ{OD WӯÐQJ Vӱ GөQJ SKѭѫQJ SKiS 1ҳP EҳW [X KѭӟQJ 7UHQG )ROORZLQJ  YjP{KuQKWKӕQJNrÿӇÿҫXWѭFKӭQJNKRiQYjF{QJFөSKiLVLQK

+ -LP 6LPRQV {QJ Oj ³ÐQJ YXD ÿӏQK OѭӧQJ´ QKj WRiQ KӑF Yj Oj QJѭӡLViQJOұSTXӻ5HQDLVVDQFH7HFKQRORJLHVWӹĈ{OD- PӝWTXӻÿҫXWѭVӱGөQJJLҧLWKXұWPi\WtQKÿӇSKkQWtFKYjÿҫXWѭFKӭQJNKRiQ

- 7UrQ WKӃ JLӟLYLӋF QJKLrQ FӭX Yj iS GөQJ FiF P{ KuQK ÿӏQK OѭӧQJ WURQJÿҫXWѭÿmSKiWWULӇQWӯUҩWOkX7KӡLJLDQJҫQÿk\YLӋFiSGөQJF{QJQJKӋWK{QJWLQWURQJÿҫXWѭEQJQәӣFiFQѭӟFFKkXÈ7X\QKLrQӣ9LӋW1DPYLӋF iS GөQJ Qj\ YүQ FKѭD SKә ELӃQ Yj KLӋQ WҥL FKӍ Fy  TXӻ Pӣ ÿӏQKOѭӧQJ9)$KRҥWÿӝQJ&yOêGRFKtQK

+ 9LӋF iS GөQJ F{QJ QJKӋ WK{QJ WLQ YjR WjL FKtQK FKӍ PӟL SKiW WULӇQ ӣ9LӋW1DPYjLQăPJҫQÿk\

+ &iFSKѭѫQJSKiS[ӃSKҥQJFәSKLӃXӣ9LӋW1DPSKҫQOӟQWKHRSKѭѫQJpháp chuyên gia

- VLӋF QJKLrQ FӭX Yj iS GөQJ F{QJ QJKӋ WK{QJ WLQ ÿӏQK OѭӧQJ WURQJ WjLFKtQKÿӇ [k\GӵQJKӋ WKӕQJ[ӃS KҥQJFә SKLӃX ÿѭӧFNǤYӑQJJL~SNKҳFSKөFFiFÿLӇP\ӃXFӫDSKѭѫQJSKiSFKX\rQJLDYjJLDWăQJKLӋXTXҧÿҫXWѭ+ӋWKӕQJGӵNLӃQVӁÿiSӭQJFiFPөFWLrXVDX

+ Dùng +ӑFPi\ [k\GӵQJP{KuQK[ӃSKҥQJFәSKLӃX

+ 6R ViQK P{ KuQK Qj\ YӟL P{ KuQK WKHR SKѭѫQJ SKiS FKX\rQ JLD ÿӇ

ÿiQKJLiPӭFÿӝKLӋXTXҧFӫDP{KuQK

2.2 Xây dӵng bài toán:

- &kX KӓL QJKLrQ FӭX WәQJ WKӇ WURQJ OXұQ YăQ Qj\ Oj OLӋX YLӋF iS GөQJ FiFWKXұW WRiQ KӑF Pi\ Fy WKӇ ÿyQJ JyS JLi WUӏ WURQJ YLӋF SKkQ ELӋW FiF FәSKLӃX KRҥW ÿӝQJ KLӋX TXҧ ÿӇ [k\ GӵQJ GDQK PөF ÿҫX Wѭ YѭӧW WUӝL KѫQ

VNIndex KRһFPӝWFKӍVӕÿҫXWѭNKiFQKѭ91+1;LQGH[ĈӇWUҧOӡLFkXKӓLQj\FiFFkXKӓLFRQVDXÿk\VӁÿѭӧFQJKLrQFӭX

+ KҧQăQJGӵÿRiQFӫDFiFWKXұWWRiQ KӑF Pi\ NKLSKkQELӋWJLӳDFiFFәSKLӃXKRҥWÿӝQJNpPKLӋXTXҧ YjKLӋXTXҧ?

+ &yWKӇVӱGөQJFiFFKLӃQOѭӧFÿҫXWѭEҵQJFiFKVӱGөQJFiFWKXұWWRiQKӑFPi\ÿӇ[k\GӵQJGDQKPөFÿҫXWѭ YѭӧWWUӝLVRYӟLWKӏWUѭӡQJKRһFPӝWFKӍVӕÿҫXWѭ?

+ &iFWKXұWWRiQKӑFPi\FyWKӇÿyQJJySYjRYLӋFOӵDFKӑQFә SKLӃXWӕWKѫQVRYӟLPӝWF{QJFөFKӑQFәSKLӃXQJүXQKLrQ"

+ &yEҵQJFKӭQJÿiQJNӇQjRYӅPӕLOLrQKӋJLӳDNKҧQăQJGӵ ÿRiQFӫDFiFWKXұWWRiQKӑFPi\ YjOӧLQKXұQGDQKPөFÿҫXWѭÿѭӧFWҥRWӯYLӋFGӵEiRQj\?

2.3 Phҥm vi và hҥn chӃ:

- 0һFGFyUҩWQKLӅXFiFKYjSKѭѫQJSKiSFyWKӇÿѭӧFVӱGөQJÿӇWUҧOӡLFiFFkXKӓLÿѭӧFQJKLrQFӭXWURQJ OXұQYăQ này, tuy nhiên, tôi xin phép ÿѭӧF[iFÿӏQKU} PӝWVӕJLӟLKҥQ QKҩWÿӏQKWURQJSKҥPYLFӫDPӝW OXұQYăQ WKҥFVƭ

2.3.1 Thuұt toán hӑc tұp:

- TroQJWKӃJLӟLFӫDKӑFPi\FyUҩWQKLӅXWKXұWWRiQKӑF máy FyWKӇÿѭӧFVӱGөQJÿӇWUҧOӡLYҩQÿӅWURQJOXұQYăQ Qj\0ӝWOӵDFKӑQWURQJVӕchúng ÿѭӧFQJKLrQFӭXWURQJOXұQYăQ này:

2.3.3 VNindex và HNXindex:

- 'DQK PөF ÿҫX Wѭ Wӯ FiF WKXұW WRiQ KӑF Pi\ Yj GDQK PөF WKDP FKLӃX VӁÿѭӧF [k\ GӵQJ Wӯ các Fә SKLӃX trong 2 FKӍ Vӕ WKӏ WUѭӡQJ FKӭQJ NKRiQchính FӫD9LӋW1DPOj 91,QGH[FӫD6ӣJLDRGӏFKFKӭQJNKRiQWKjQKSKӕ

+ӗ&Kt0LQKYj+1;LQGH[FӫD6ӣJLDRGӏFKFKӭQJNKRiQ+j1ӝL, trong JLDLÿRҥQWӯWKiQJQăPÿӃQWKiQJQăP

2.3.4 ChiӃQOѭӧFÿҫu Wѭ:

- CKLӃQOѭӧFÿҫXWѭÿѭӧFÿiQKJLiWURQJOXұQYăQ Qj\FKӍJLӟLKҥQӣFKLӃQOѭӧF³0XDYjQҳPJLӳ´ĈLӅXQj\GүQÿӃQFKLӃQOѭӧFFKӍÿѭӧFKѭӣQJOӧLWӯYLӋFWăQJJLiYjNK{QJWKӇWKXÿѭӧFOӧLQKXұQWӯYLӋFJLҧP giá FәSKLӃX

Ѭu ÿLӇPFӫDFiFFKLӃQOѭӧFQj\OjGӉiSGөQJ trong FҧOêWKX\ӃWYjWKӵFWӃ

2.4 Cҩu trúc:

- /XұQYăQ ÿѭӧFFҩXWU~FWKjQKEҧ\ FKѭѫQJ&KѭѫQJÿҫXWLrQOjWyPWҳWQӝLGөQJFӫDOXұQYăQ&KѭѫQJWKӭKDLYjKLӋQWҥL, là JLӟLWKLӋX[k\GӵQJYҩQÿӅSKҥPYL JLӟLKҥQYjFҩXWU~F3KҫQFzQOҥLFӫDOXұQYăQ Qj\ÿѭӧFWәFKӭFQKѭVDX

+ &KѭѫQJ ± 3KѭѫQJSKiS[ӃSKҥQJFәSKLӃXWURQJÿҫXWѭFKӭQJNKRiQ

&KѭѫQJQj\EҳWÿҫXYӟLPӝWSKҫQWyPWҳW YjWKҧROXұQFӫDJLҧL1REHONLQKWӃQăPYӅYLӋFGӵEiRJLiFKӭQJNKoán3KҫQWLӃSWKHRP{Wҧ SKѭѫQJ SKiS SKkQ WtFK Fѫ EҧQ WURQJ YLӋF SKkQ WtFK Gӵ ÿRiQ FKӭQJkhoán 3KҫQ WKӭ ED JLӟL WKLӋX YӅ FiF SKѭѫQJ SKiS FKX\rQ JLD WURQJYLӋF SKkQ WtFK Yj [ӃS KҥQJ Fә SKLӃX &KѭѫQJ Qj\ NӃW WK~F YӟL SKҫQxem xét các EjLQJKLrQFӭX OLrQTXDQYӅӭQJGөQJKӑFPi\WURQJOӵDFKӑQFәSKLӃXYjGӵÿRiQ

+ &KѭѫQJ - +ӑFPi\

&KѭѫQJQj\NKiPSKiOêWKX\ӃWYӅKӑFPi\ÿѭӧFiSGөQJWURQJOXұQYăQ3KҫQÿҫXWLrQJLӟLWKLӋXêWѭӣQJFKXQJYӅKӑFPi\EҵQJFiFKP{WҧFiFORҥLYҩQÿӅKӑFWұSNKiFQKDXYjJLҧLWKtFKYҩQÿӅOLrQTXDQÿӃQoverfitting3KҫQWKӭKDLWKҧROXұQYӅFiFWKXұWWRiQKӑFPi\ VӁÿѭӧFQJKLrQ FӭX WURQJ OXұQ YăQ này 3KҫQ FXӕL FQJ P{ Wҧ FiF YҩQ ÿӅWKѭӡQJ JһS Yj FiF SKѭѫQJ SKiS [ӱ Oê WURQJ TXi WUuQK WLӅQ [ӱ Oê GӳOLӋXÿiQKJLiYjOӵDFKӑQP{KuQKSKKӧS

+ &KѭѫQJ - 3KѭѫQJSKiSOXұQ

&KѭѫQJQj\WұSWUXQJYjRWҩWFҧFiFSKѭѫQJSKiSÿѭӧFQJKLrQFӭXÿӇWUҧOӡLYҩQÿӅWURQJOXұQYăQ Qj\7KӭQKҩWFiFSKѭѫQJSKiS[k\GӵQJGDQKPөF ÿҫXWѭYjOҩ\GӳOLӋXFҫQWKLӃWÿѭӧFWUuQKEj\7KӭKDL, quá trình gán nhãn ÿӇSKkQELӋWJLӳDFiFFәSKLӃXKLӋXTXҧYjNpPKLӋXTXҧÿѭӧFJLӟLWKLӋX&KѭѫQJQj\FNJQJWUuQKEj\FiFKWKӭF ÿӇVӱGөQJFiFWKXұWWRiQKӑFPi\ ÿӇ[ӃSKҥQJOӵDFKӑQ FәSKLӃXYj[k\GӵQJGDQKPөFÿҫXWѭ&iFVӕOLӋXÿѭӧFVӱGөQJÿӇÿiQK JLiKLӋXVXҩWFӫDGDQKPөFÿҫXWѭÿѭӧFWUuQKEj\YjFXӕLFQJOjVӵOӵDFKӑQGDQKPөFWKDPFKLӃXÿѭӧFWKҧROXұQ

+ &KѭѫQJ - 7KtQJKLӋP .ӃWTXҧ

&KѭѫQJQj\FXQJFҩS các NӃWTXҧWKtQJKLӋPiSGөQJWUrQ GӳOLӋXOӏFh VӱFKRWӯQJWKXұWWRiQKӑFPi\&iFNӃW TXҧYӅ KLӋXVXҩWGӵÿRiQVӁÿѭӧFSKkQWtFK FKRWӯQJWKXұWWRiQYjVRViQKFK~QJYӟLQKDXÿӇWuPUDÿѭӧFJLҧLWKXұWSKKӧSQKҩWFKRYLӋF[ӃSKҥQJFәSKLӃXYjEӝGӳOLӋXFӫD WKӏ WUѭӡQJ FKӭQJ NKRiQ 9LӋW 1DP *LҧL WKXұW Qj\ VDX ÿy ÿѭӧFGQJÿӇWKLӃWOұSGDQK PөFÿҫXWѭYjVRViQKYӟLGDQKPөFWKDPFKLӃX ÿӇNKiPSKiFiFFkXKӓLFRQÿѭӧFQJKLrQFӭXWURQJOXұQYăQ

+ &KѭѫQJ - ӃWOXұQ

&KѭѫQJQj\WyPWҳWOҥLWRjQEӝOXұQYăQEҵQJFiFKWUҧOӡLFiFFkXKӓLFKtQKYjSKөÿѭӧF[k\GӵQJWUѭӟFÿy3KҫQWLӃSWKHRFKӍUDFiFKҥQFKӃFӫDOXұQYăQWӯÿyÿѭDUDKѭӟQJSKiWWULӇQWLӃSWKHR

3 3KѭѫQJ3KiS;ӃS+ҥQJ&ә3KLӃX 7URQJĈҫX7ѭ&KӭQJ.KRiQ:

3.1 Dӵ báo giá chӭng khoán:

- K{QJ Fy FiFK QjR Gӵ ÿRiQ ÿѭӧF JLi Fә SKLӃX FKӭQJ NKRiQ WURQJ YjLQJj\KRһF YjLWXҫQWӟL7X\QKLrQYLӋFGӵÿRiQӣWҫPGjLKҥQKѫQOjÿLӅXFyWKӇOjPÿѭӧFYtGөGӵÿRiQJLiWURQJEDKD\QăPQăPWӟL9LӋQ+jQOkP.KRDKӑF7Kө\ĈLӇQWuPWKҩ\NӃWOXұQWUrQ- YӕQUҩWÿiQJQJҥFQKLrQYj QJKH Fy Yҿ PkX WKXүQ - WURQJ QJKLrQ FӭX FӫD ED QKj NKRD KӑFLaureates, Eugene Fama, Lars Peter Hansen và Robert Shiller

- %ҳW ÿҫX Wӯ QKӳQJ QăP  (XJHQH )DPD Yj PӝW Vӕ FӝQJ Vӵ ÿm FKӭQJPLQKUҵQJJLiFKӭQJNKRiQUҩWNKyÿRiQWURQJQJҳQKҥQYjUҵQJWLQWӭFPӟLFyҧQKKѭӣQJFӵFNǤQKDQKFKyQJWӟLJLiFҧ1KӳQJQJKLrQFӭXQj\NK{QJ FKӍ OjP WiF ÿӝQJ QӅQ ÿӃQ QKӳQJ QJKLrQ FӭX YӅ VDX Pj FzQ FyQKӳQJ WKD\ ÿәL ÿiQJ NӇ WӟL WKӏ WUѭӡQJ 1KӳQJ TXӻ ÿҫX Wѭ WKHR FKӍ Vӕ,QGH[)XQG ÿDQJQJj\PӝWQӣUӝWUrQWKӏWUѭӡQJFKӭQJNKRiQWRjQFҫXKLӋQQD\OjPӝWWURQJQKӳQJ YtGөWLrXELӇX

- 1ӃXJLiFҧKҫXQKѭNK{QJWKӇÿRiQÿѭӧFWURQJSKҥPYLYjLQJj\KD\YjLWXҫQOLӋXFK~QJFyWUӣQrQFjQJNKyÿRiQWURQJYzQJYjLQăP&kXWUҧOӡLOjNK{QJQKѭ5REHUW6KLOOHUWӯQJNKiPSKiUDYjRÿҫXQKӳQJQăPÐQJ SKiW KLӋQ UD UҵQJ JLi FKӭQJ NKRiQ GDR ÿӝQJ QKLӅX KѫQ Fә WӭF FiFF{QJW\YjUҵQJWӹOӋJLӳDJLiYjFәWӭFFy[XKѭӟQJÿL[XӕQJNKLFәWӭFFDRYjFy[XKѭӟQJWăQJNKLFәWӭFJLҧP&{QJWKӭFQj\NK{QJFKӍÿ~QJYӟLFKӭQJNKRiQPjFzQÿ~QJYӟLWUiLSKLӃXYjFiFORҥLWjLVҧQNKic

- /ӧLQKXұQFDRWURQJWѭѫQJODLÿѭӧF[HPOjNKRҧQEÿҳSFKRYLӋFQҳPJLӳWjLVҧQUӫLURWURQJQKӳQJWKӡLÿLӇPUӫLUREҩWWKѭӡQJ1KjNKRDKӑFWKӭED WURQJ *LҧL WKѭӣQJ OҫQ Qj\ {QJ /DUV 3HWHU +DQVHQ ÿm SKiW WULӇQ PӝWSKѭѫQJSKiSWKӕQJNrFyWKӇSKKӧSYӟLYLӋFWKӱQJKLӋPF{QJWKӭFWӹOӋWUrQYjRYLӋFÿӏQKJLiWjLVҧQWUrQWKӵFWӃ

3.2 3KkQWtFKFѫEҧn WURQJÿҫXWѭFKӭng khoán:

- 3KkQWtFKFѫEҧQOjPӝWFKLӃQOѭӧFSKkQWtFKÿҫXWѭFәSKLӃXKRһFFKӭQJNKRiQEҵQJFiFK[iFÿӏQKJLiWUӏQӝLWҥLFӫDQy0ӝWWKjQKSKҫQUҩWTXDQWUӑQJFӫDSKѭѫQJSKiSQj\OjSKҧL[HP[pWWuQKWUҥQJWjLFKtQKFӫD PӝWF{QJW\&iFNKtDFҥQKNKiFQKѭTXҧQOê[XKѭӟQJF{QJQJKLӋSYjÿLӅXNLӋQWәQJWKӇFӫDQӅQNLQKWӃFNJQJÿѭӧFWtQKÿӃQ0өFWLrXFKtQKOjѭӟFWtQKPӝWJLiWUӏQKҩWÿӏQK FKRF{QJW\ÿӇFyWKӇÿѭӧFVӱGөQJOjPFѫVӣTX\ӃWÿӏQK1ӃXGӳOLӋXYjWK{QJWLQKѭӟQJWӟLJLiWUӏFDRKѫQJLiWUӏKLӋQWҥLÿDQJÿѭӧFÿѭDUDWUrQWKӏWUѭӡQJJLiWUӏKLӋQWҥLFӫDFәSKLӃXÿѭӧFFRLOjEӏÿӏQKJLiWKҩS1yLFiFKNKiFFiFQKjÿҫXWѭFyWKӇ WKXOӡLWӯNKRҧQJWUӕQJSKiWWULӇQWURQJJLiWUӏFӫDQy1JѭӧFOҥLQӃXJLiWUӏWKҩSKѫQVRYӟLJLiQJҳQKҥQKLӋQWҥLF{QJW\ÿѭӧFFRLOjÿӏQKJLiTXiFDRYjJLiVӁFy[XKѭӟQJJLҧPGjLKҥQ

- 7URQJSKkQWtFKFѫEҧQELӋQSKiSFKӫ\ӃXÿѭӧFFiFQKjSKkQWtFKVӱGөQJÿӇOӵDFKӑQÿҫXWѭYjRFәSKLӃXOjFiFQJX\rQWҳFFѫEҧQFӫDQy3KҥPYLFӫDQyNKiUӝQJYuQyEDRJӗPVӭFPҥQKWjLFKtQKOmQKÿҥRQJjQKYjYLӋF

TXҧQ Oê FKҩW OѭӧQJ WӕW 1KӳQJ QJѭӡL Vӱ GөQJ SKѭѫQJ SKiS Fѫ EҧQ WuPNLӃP QKӳQJ JLi WUӏ Eӏ ÿiQK JLi WKҩS Fy NKҧ QăQJ VLQK OӧL FDR Yj WăQJGRDQKWKXWӯFiFKRҥWÿӝQJNLQKGRDQKFӕWO}LFyNKҧQăQJWUҧQӧYjFyPӝWGzQJWLӅQWӵGRÿiQJNӇEDRJӗPNKҧQăQJÿѭDYjRVӱGөQJKLӋXTXҧ

&iFFKӍVӕWjLFKtQKPjFiFQKjÿҫXWѭTXDQWkPEDRJӗPWKXQKұSWUrQPӛLFәSKLӃX(36 KӋVӕWKӏJLiYjWKXQKұSFәSKLӃX3( WӹOӋJLiKD\JLiWUӏJKLVәWӹVӕ3%  Oj KjQJQJjQFһSҧQKPһWQJѭӡL YjҧQKNK{QJSKҧLPһWQJѭӡL ÿѭӧFÿѭDYjR

&K~êOjGӳOLӋXQj\FKӍSKkQELӋWPһWQJѭӡLYjNK{QJSKҧLPһWQJѭӡLPjNK{QJSKkQELӋWNKX{QPһWFӫDQKӳQJQJѭӡLNKiFQKDX

7KXұWWRiQ+ӑFFyJLiPViW FzQÿѭӧFWLӃSWөFFKLDQKӓUDWKjQKKDLORҥLchính:

+ &ODVVLILFDWLRQ3KkQORҥL

MӝWEjLWRiQÿѭӧc gӑi là Phân loҥi nӃu các nhãn cӫa dӳ liӋXÿҫu vào ÿѭӧc chia thành mӝt sӕ hӳu hҥn nhóm Ví dө*PDLO[iFÿӏnh xem mӝt email có phҧi là spam hay không; các hãng tín dөQJ[iFÿӏnh xem mӝt khách hàng có khҧ QăQJWKDQKWRiQQӧ hay không Ba ví dө phía trên ÿѭӧc chia vào loҥi này

+ RegressiRQ+ӗLTX\

NӃu nhãn NK{QJ ÿѭӧc chia thành các nhóm mà là mӝt giá trӏ thӵc cө thӇ Ví dө: mӝWFăQQKjUӝng x m2, có y phòng ngӫ và cách trung tâm thành phӕ z km sӁ có giá là bao nhiêu?

GҫQÿk\0LFURVRIWFyPӝt ӭng dөng dӵ ÿRiQJLӟi tính và tuәi dӵa trên khuôn mһt Phҫn dӵ ÿRiQ JLӟi tính có thӇ coi là thuұt toán Phân loҥi, phҫn dӵ ÿRiQWXәi có thӇ coi là thuұt toán Hӗi quy Chú ý rҵng phҫn dӵ ÿRiQWXәLFNJQJFyWKӇ coi là Phân loҥi nӃu ta coi tuәi là mӝt sӕ nguyên GѭѫQJNK{QJOӟQKѫQFK~QJWDVӁ có 150 class (lӟp) khác nhau

- 8QVXSHUYLVHG/HDUQLQJ+ӑFNK{QJJLiPViW

7URQJWKXұWWRiQQj\FK~QJWDNK{QJELӃWÿѭӧFNӃWTXҧ KD\QKmQPjFKӍFyGӳOLӋXÿҫXYjR7KXұWWRiQKӑFNK{QJJLiPViW VӁGӵDYjRFҩXWU~FFӫDGӳ OLӋX ÿӇ WKӵF KLӋQ PӝW F{QJ YLӋF QjR ÿy Yt Gө QKѭ SKkQ QKyPFOXVWHULQJ KRһFJLҧPVӕFKLӅXFӫDGӳOLӋXGLPHQVLRQUHGXFWLRQ ÿӇWKXұQWLӋQWURQJYLӋFOѭXWUӳYjWtQKWRiQ

0ӝWFiFKWRiQKӑF+ӑFNK{QJJLiPViW là khLFK~QJWDFKӍFyGӳOLӋXYjR

;PjNK{QJELӃWQKmQ<WѭѫQJӭQJ

1KӳQJWKXұWWRiQORҥLQj\ÿѭӧFJӑLOj+ӑFNK{QJJLiPViW YuNK{QJJLӕQJQKѭ+ӑFFyJLiPViWFK~QJWDNK{QJELӃWFkXWUҧOӡLFKtQK[iFFKRPӛLGӳOLӋXÿҫXYjR*LӕQJQKѭNKLWDKӑFNK{QJFyWKҫ\F{JLiRQjRFKӍFKRWDELӃWÿyOjFKӳ$KD\FKӳ%&өPNK{QJJLiPViWÿѭӧFÿһW WrQWKHRQJKƭDnày

Các bài toán +ӑFNK{QJJLiPViW ÿѭӧFWLӃSWөFFKLDQKӓWKjQKKDLORҥL + Clustering (phân nhóm)

Mӝt bài toán phân nhóm toàn bӝ dӳ liӋu X thành các nhóm nhӓ dӵa trên sӵ liên quan giӳa các dӳ liӋu trong mӛi nhóm Ví dө: phân nhóm khách hàng dӵDWUrQKjQKYLPXDKjQJĈLӅXQj\FNJQJJLӕQJQKѭYLӋc WDÿѭDFKRPӝWÿӭa trҿ rҩt nhiӅu mҧnh ghép vӟi các hình thù và màu sҳc khác nhau, ví dө tam giác, vuông, tròn vӟLPjX[DQKYjÿӓVDXÿyyêu cҫu trҿ phân chúng thành tӯng nhóm Mһc dù không cho trҿ biӃt mҧQKQjRWѭѫQJӭng vӟi hình nào hoһc màu nào, nhiӅu khҧ QăQJFK~QJvүn có thӇ phân loҥi các mҧnh ghép theo màu hoһc hình dҥng

+ Association

Là bài toán khi chúng ta muӕn khám phá ra mӝt quy luұt dӵa trên nhiӅu dӳ liӋX FKR WUѭӟc Ví dө: nhӳng khách hàng nam mua quҫn áo WKѭӡQJFy[XKѭӟQJPXDWKrPÿӗng hӗ hoһc thҳWOѭQJQKӳng khán giҧ [HPSKLP6SLGHU0DQWKѭӡQJFy[XKѭӟng xem thêm phim Bat Man,

dӵD YjR ÿy Wҥo ra mӝt hӋ thӕng gӧi ý khách hàng (Recommendation 6\VWHP WK~Fÿҭy nhu cҫu mua sҳm

- Semi-SuSHUYLVHG/HDUQLQJ+ӑFEiQJLiPViW

Các bài toán khi chúnJWDFyPӝWOѭӧQJOӟQGӳOLӋX; QKѭQJFKӍPӝWSKҫQWURQJFK~QJÿѭӧFJiQQKmQÿѭӧFJӑLOj+ӑFEiQJLiPViW1KӳQJEjLWRiQWKXӝFQKyPQj\QҵPJLӳDKDLQKyPÿѭӧFQrXErQWUrQ

0ӝWYtGөÿLӇQKuQKFӫDQKyPQj\OjFKӍFyPӝWSKҫQҧQKKRһFYăQEҧQÿѭӧFJiQQKmQYtGөEӭFҧQKYӅQJѭӡLÿӝQJYұWKRһFFiFYăQEҧQNKRDKӑFFKtQKWUӏ YjSKҫQOӟQFiFEӭFҧQKYăQEҧQNKiFFKѭDÿѭӧFJiQQKmQÿѭӧFWKXWKұSWӯLQWHUQHW7KӵFWӃFKRWKҩ\UҩWQKLӅXFiFEjLWRiQ+ӑFPi\ WKXӝF YjR QKyP Qj\ Yu YLӋF WKX WKұS Gӳ OLӋX Fy QKmQ WӕQ UҩW QKLӅX WKӡLJLDQYjFyFKLSKtFDR5ҩWQKLӅXORҥLGӳOLӋXWKұPFKtFҫQSKҧLFyFKX\rQJLDPӟLJiQQKmQÿѭӧFҧQK\KӑFFKҷQJKҥQ 1JѭӧFOҥLGӳOLӋXFKѭDFynhãn có WKӇÿѭӧFWKXWKұSYӟLFKLSKtWKҩSWӯLQWHUQHW

- 5HLQIRUFHPHQW/HDUQLQJ+ӑF&ӫQJ&ӕ

+ӑFFӫQJFӕ OjFiFEjLWRiQJL~SFKRPӝWKӋWKӕQJWӵÿӝQJ[iFÿӏQKKjQKYLGӵDWUrQKRjQFҧQKÿӇÿҥWÿѭӧFOӧLtFKFDRQKҩW+LӋQWҥLKӑFFӫQJFӕ FKӫ \ӃX ÿѭӧF iS GөQJ YjR /ê 7KX\ӃW 7Uz &KѫL FiF WKXұW WRiQ FҫQ [iFÿӏQKQѭӟFÿLWLӃSWKHRÿӇÿҥWÿѭӧFÿLӇPVӕFDRQKҩW

4.1.2 Hàm mҩt mát và tham sӕ mô hình:

- 0ӛL P{ KuQK KӑF Pi\ ÿѭӧF P{ Wҧ EӣL FiF WKDP Vӕ P{ KuQK PRGHOSDUDPHWHUV &{QJYLӋFFӫDPӝWWKXұWWRiQKӑFPi\OjÿLWuPFiFWKDPVӕP{KuQKSKKӧSYӟLPӛLEjLWRiQ9LӋFÿLWuPWKDPVӕP{KuQKFyOLrQTXDQPұWWKLӃWÿӃQFiFSKpSÿiQKJLi0өFÿtFKFӫDFK~QJWDOjÿLWuPFiFWKDPVӕP{KuQKVDRFKRFiFSKpSÿiQKJLiFKRNӃWTXҧWӕWQKҩW7URQJEjLWRiQSKkQOӟSNӃWTXҧWӕWFyWKӇÿѭӧFKLӇXOjtWÿLӇPGӳOLӋXÿѭӧFSKkQOӟSVDLQKҩW7URQJEjLWRiQKӗLTX\NӃWTXҧWӕWOjNKLVӵVDLOӋFKJLӳDÿҫXUDGӵÿRiQYjÿҫXUDWKӵFVӵOjtWQKҩW

- QXDQKӋJLӳDPӝWSKpSÿiQKJtDYj FiFWKDPVӕP{KuQKWKѭӡQJÿѭӧFP{WҧWK{QJTXDPӝWKjPVӕÿѭӧFJӑLOjKjPPҩWPiWORVVIXQFWLRQKD\FRVWIXQFWLRQ +jPPҩWPiWQj\WKѭӡQJFyJLiWUӏQKӓ NKLSKpSÿiQKJLiFKRNӃWTXҧ WӕWYjQJѭӧFOҥL9LӋF ÿLWuPFiFWKDPVӕP{KuQKVDRFKRSKpSÿiQKJLiWUҧYӅNӃWTXҧWӕWWѭѫQJÿѭѫQJYӟLYLӋFWӕLWKLӇXKjPPҩWPiW1KѭYұ\YLӋF[k\GӵQJPӝWP{KuQKKӑFPi\FKtQKOjYLӋFÿLJLҧLPӝWEjLWRiQWӕLѭX4XiWUuQKÿyFyWKӇÿѭӧFFRLOjTXiWUuQKKӑFFӫDPi\

4.1.3 Over- và underfitting

- 0ӛLNKLWKҧROXұQYӅPӝWP{KuQK GӵÿRiQÿLӅXTXDQWUӑQJQKҩWFҫQTXDQWkPOjFiFGӵÿRiQEӏVDLOӋFKVRYӟLWKӵFWӃELDVYjYDULDQFH