The Predictability of Tree-based Machine Learning Algorithms in the Big Data Context

Abstract

This research work is concerned with the predictability of ensemble and singular tree-based machine learning algorithms during the recession and prosperity of the two companies listed in the Tehran Stock Exchange in the context of big data. In this regard, the main issue is that economic managers and the academic community require predicting models with more accuracy and reduced execution time; moreover, the prediction of the companies recession in the stock market is highly significant. Machine learning algorithms must be able to appropriately predict the stock return sign during the market downturn and boom days. Addressing the stated challenge will upgrade the quality of stock purchases and, subsequently, will increase profitability. In this article, the proposed solution relies on the utilization of tree-based machine learning algorithms in the context of big data. The proposed solution exploits the decision tree algorithm, which is a traditional and singular tree-based learning algorithm. Furthermore, two modern and ensemble tree-based learning algorithms, random forest and gradient boosted tree, has been utilized for predicting the stock return sign during recession and prosperity. The mentioned cases were implemented by applying the machine learning tools in python programming language and PYSPARK library that is used explicitly for the big data context. The utilized research data of the current study are the shares information of two companies of the Tehran Stock Exchange. The obtained results reveal that the applied ensemble learning algorithms have performed better than the singular learning algorithms. Additionally, adding 23 technical features to the initial data and subsequent applying of the PCA feature reduction method have demonstrated the best performance among other modes. In the meantime, it has been concluded that the initial data do not possess the proper resolution or generalizability, either during prosperity or recession.

Decision Fusion for Stock Market Prediction: A Systematic Review

“…The second most popular method for fusing classification results is the tree-based method, which involves first feeding forecasts of all base learners into a tree-based algorithm, then mapping each prediction to a neighborhood in the set of dependent variables, and then returning the mean neighborhood [99]. The most commonly used tree-based methods include gradient boosting [40], [75], [96] and random forest [28], [40], [75]. It is worth noting that Barak et al [27] used five tree-based methods for decision fusion: the BF tree, decision table, decision tree, decision tree naïve Bayes (DTNB), and the LAD tree, with the decision table performing the best.…”

“…Author(s) ANN [22], [33], [41], [42], [57], [58] Decision Tree [28], [40], [75], [96] SVM [43], [68], [93] LSTM [30], [88] PNN [34] ELM [48] DBN [84]…”

“…In addition, the final forecast can be the fusion result of different types of predictions, such as the fusion of stock movement trends and external factors [56], the fusion of stock prices and external factors [51], the fusion of stock movement probability and external factors [78], and the fusion of stock volatility and external factors [50]. Binary Stock movement trend (e.g., up and down) [28], [29], [32], [39], [40], [41], [42], [43], [44], [46], [48], [62], [64], [68], [73], [75], [77], [84], [85], [88] Stock movement probability [30], [49], [78], [93], [96] Multiclass Stock movement trend (e.g., up, neutral, and down) [22], [34], [56] Stock turning indicator [57], [58] Stock risk level [27] Confidence level of stock movement [33] Effect level of external factors [56] TABLE VIII STUDIES RELATED TO REGRESSION TASKS Forecasts of Base Learners Author(s)…”

The Construction of Scalable Decision Tree based on Fast Splitting and J-Max Pre-Pruning on Large Datasets

“…By increasing the complexity of the decision tree, a considerable increase takes place in the occurrence probability of overfitting. In addition, the training error decreases while the test error increases [2]. The reason for the occurrence of this phenomenon is the noise in the training dataset or inappropriate selection of training data.…”

“…Classification is one of the most widely used methods for data mining in order to provide a model for specifying the label of different samples based on their characteristics. In this regard, the decision tree is one of the most widely used algorithms which can produce understandable human descriptions of relationships in a dataset [2]. Further, this algorithm is one of the most widely used algorithms in pattern recognition domain due to its simplicity and interpretation, rule representation in a hierarchical format, cost and time of proper construction, the ability to work with continuous and discrete data, the need for prior knowledge and accurate presentation.…”