This paper examines split-improvement feature importance scores for tree-based methods. Starting with Classification and Regression Trees (CART; Breiman, 2017) and C4.5 (Quinlan, 2014), decision trees have been a workhorse of general machine learning, particularly within ensemble methods such as Random Forests (RF; Breiman, 2001) and Gradient Boosting Trees (Friedman, 2001). They enjoy the benefits of computational speed, few tuning parameters and natural ways of handling missing values.

Random forests (RF) are one of the most widely used ensemble learning methods in classification and regression tasks. Despite its impressive performance, its theoretical consistency, which would ensure that its result converges to the optimum as the sample size increases, has been left far behind. Several consistent random forest variants have been proposed, yet all with relatively poor performance compared to the original random forests. In this paper, a novel RF framework named multinomial random forests (MRF) is proposed. In the MRF, an impurity-based multinomial distribution is constructed as the basis for the selection of a splitting point. This ensures that a certain degree of randomness is achieved while the overall quality of the trees is not much different from the original random forests. We prove the consistency of the MRF and demonstrate with multiple datasets that it performs similarly as the original random forests and better than existent consistent random forest variants for both classification and regression tasks.

Stochastic partition models divide a multi-dimensional space into a number of rectangular regions, such that the data within each region exhibit certain types of homogeneity. Due to the nature of their partition strategy, existing partition models may create many unnecessary divisions in sparse regions when trying to describe data in dense regions. To avoid this problem we introduce a new parsimonious partition model -- the Rectangular Bounding Process (RBP) -- to efficiently partition multi-dimensional spaces, by employing a bounding strategy to enclose data points within rectangular bounding boxes. Unlike existing approaches, the RBP possesses several attractive theoretical properties that make it a powerful nonparametric partition prior on a hypercube. In particular, the RBP is self-consistent and as such can be directly extended from a finite hypercube to infinite (unbounded) space. We apply the RBP to regression trees and relational models as a flexible partition prior. The experimental results validate the merit of the RBP {in rich yet parsimonious expressiveness} compared to the state-of-the-art methods.

This study investigates the potential effects of different Dynamic Message Signs (DMSs) on driver behavior using a full-scale high-fidelity driving simulator. Different DMSs are categorized by their content, structure, and type of messages. A random forest algorithm is used for three separate behavioral analyses; a route diversion analysis, a route choice analysis and a compliance analysis; to identify the potential and relative influences of different DMSs on these aspects of driver behavior. A total of 390 simulation runs are conducted using a sample of 65 participants from diverse socioeconomic backgrounds. Results obtained suggest that DMSs displaying lane closure and delay information with advisory messages are most influential with regards to diversion while color-coded DMSs and DMSs with avoid route advice are the top contributors impacting route choice decisions and DMS compliance. In this first-of-a-kind study, based on the responses to the pre and post simulation surveys as well as results obtained from the analysis of driving-simulation-session data, the authors found that color-blind-friendly, color-coded DMSs are more effective than alphanumeric DMSs - especially in scenarios that demand high compliance from drivers. The increased effectiveness may be attributed to reduced comprehension time and ease with which such DMSs are understood by a greater percentage of road users.

We present a visual symptom checker that combines a pre-trained Convolutional Neural Network (CNN) with a Reinforcement Learning (RL) agent as a Question Answering (QA) model. This method enables us to not only increase the classification confidence and accuracy of the visual symptom checker, but also decreases the average number of relevant questions asked to narrow down the differential diagnosis. By combining the CNN output in the form of classification probabilities as a part of the state structure of the simulated patient's environment, a DQN-based RL agent learns to ask the best symptom that maximizes its expected return over symptoms. We demonstrate that our RL approach increases the accuracy more than 20% as compared to the CNN alone, and up to 10% as compared to the decision tree model. We finally show that the RL approach not only outperforms the performance of the decision tree approach but also narrows down the diagonosis faster in terms of the average number of asked questions.

You may have thought we were done with decisions trees. I am done with respect to discussing general approaches and types of problems. You could say that we're moving from a view of the forest, to finding the root for our tree. However, there is a bit more to explore when it comes to the underlying mathematical functions associated with navigating data to construct our trees. In our last discussion, I introduced the concept of a cost function and gave a specific example in the Gini coefficient.

Has this happened to you? You are working on your dataset. You create a classification model and get 90% accuracy immediately. You dive a little deeper and discover that 90% of the data belongs to one class. This is an example of an imbalanced dataset and the frustrating results it can cause.

Has this happened to you? You are working on your dataset. You create a classification model and get 90% accuracy immediately. You dive a little deeper and discover that 90% of the data belongs to one class. This is an example of an imbalanced dataset and the frustrating results it can cause.

We present a new distributed fuzzy partitioning method to reduce the complexity of multi-way fuzzy decision trees in Big Data classification problems. The proposed algorithm builds a fixed number of fuzzy sets for all variables and adjusts their shape and position to the real distribution of training data. A two-step process is applied : 1) transformation of the original distribution into a standard uniform distribution by means of the probability integral transform. Since the original distribution is generally unknown, the cumulative distribution function is approximated by computing the q-quantiles of the training set; 2) construction of a Ruspini strong fuzzy partition in the transformed attribute space using a fixed number of equally distributed triangular membership functions. Despite the aforementioned transformation, the definition of every fuzzy set in the original space can be recovered by applying the inverse cumulative distribution function (also known as quantile function). The experimental results reveal that the proposed methodology allows the state-of-the-art multi-way fuzzy decision tree (FMDT) induction algorithm to maintain classification accuracy with up to 6 million fewer leaves.

The performance of classification algorithms with a massive and highly imbalanced data stream depends upon efficient balancing strategy. Some techniques of balancing strategy have been applied in the past with Batch data to resolve the class imbalance problem. This paper proposes a new incremental data balancing framework which can work with massive imbalanced data streams. In this paper, we choose Racing Algorithm as an automated data balancing technique which optimizes the balancing techniques. We applied Random Forest classification algorithm which can deal with the massive data stream. We investigated the suitability of Racing Algorithm and Random Forest in the proposed framework. Applying new technique in the proposed framework on the European Credit Card dataset, provided better results than the Batch mode. The proposed framework is more scalable to handle online massive data streams.