Title Page
Contents
Abstract 7
1. Introduction 8
2. Background and literature review 10
2.1. Prediction of stock price direction: using sentimental analysis 10
2.2. Prediction of stock price direction: using machine learning and deep learning 12
3. Research methodology 15
3.1. Research design 15
3.2. Sample selection 17
3.3. Classifiers 20
3.3.1. Logistic Regression (LR) 20
3.3.2. Neural Network (NN) 20
3.3.3. Decision Tree (DT) 21
3.3.4. Support Vector Machine (SVM) 21
3.3.5. Random Forest (RF) 21
3.3.6. Bagging 21
4. Proposed framework 22
4.1. Convolutional Neural Network (CNN) 22
4.2. Model estimation 24
5. Results and discussions 26
5.1. Accuracy Performance Comparison Results 26
5.2. Precision Performance Comparison Results 27
5.3. Recall Performance Comparison Results 28
5.4. F1-Score (F-measure) Performance Comparison Results 29
5.5. Additional Test Result 30
6. Conclusions, limitations, and future research directions 31
Reference 34
〈Table 1〉 Prior literatures related to prediction of stock price direction: using sentimental analysis 11
〈Table 2〉 Prior literatures related to prediction of stock price direction: using machine learning 13
〈Table 3〉 Prior literatures related to prediction of stock price direction: using deep learning 14
〈Table 4〉 Description of audit report 18
〈Table 5〉 Example of sentiment words in the audit report 19
〈Table 6〉 Description of dependent variable 20
〈Table 7〉 Confusion matrix of classification 26
〈Table 8〉 Result of Stock price direction by Accuracy 27
〈Table 9〉 Result of Stock price direction by Precision 28
〈Table 10〉 Result of Stock price direction by Recall 29
〈Table 11〉 Result of Stock price direction by F1-Score 30
〈Table 12〉 Result of the highest performance of Stock price prediction direction 31
〈Figure 1〉 Procedure of this research 17
〈Figure 2〉 Architecture of LeNet-5, a Convolution Neural Net 22
〈Figure 3〉 2x2 convolution filter 23
〈Figure 4〉 Max pooling filter 23
〈Figure 5〉 The Architecture of Convolutional neural network proposed in this study 24