business data teams

Abstract: In the age of data-driven decision-making, deep learning models have become powerful tools for uncovering patterns and making predictions that influence many aspects of society. However, their broader adoption in scientific and business applications is often limited by a lack of interpretability. Feature weighting methods, which assess the relevance of individual features, are among the most widely studied approaches for improving interpretability. While post hoc techniques have advanced significantly, especially in modeling nuanced relationships between features and labels in multiclass problems, embedded methods typically fall short in this regard. Embedded techniques integrate feature relevance learning directly into the model training process, but they are often unable to capture label-specific relationships. To address the limitation, we propose multiple extensions to the Feature Selection Layer, an embedded feature weighting mechanism for neural networks. Through extensive experimentation, we identify the most effective extension and benchmark it against established feature weighting methods. Our results show that the proposed method matches existing techniques in identifying globally relevant features and outperforms them in capturing label-specific relevance. By leveraging the internal dynamics of the trained neural network, it reveals how individual features contribute to the prediction of specific labels, requiring no post-processing beyond the training process.

Abstract: Class-imbalance methods implicitly assume that the minority class is uniformly undersampled relative to the majority class. However, in many real-world settings, minority instances may be disproportionately under-observed in certain regions of the feature space. For example, small businesses that go bankrupt may disappear from records, while those that survive remain visible, making bankruptcy appear less common among small firms than it actually is. This gives rise to localized minority imbalance (LMI), a challenge that is often overlooked and extends beyond general class-count imbalance. We show that under LMI, existing imbalance mitigation techniques can fit observation-induced biases in the training data and generalize poorly to under-observed regions of the true minority distribution. To address this, we propose a tree-based stratified approach that recursively partitions the feature space with the goal of reducing within-stratum LMI distortion. For each resulting stratum, we pair its majority instances with the full observed minority set and train a base classifier to create an ensemble. Extensive experiments over benchmark tabular datasets simulated with LMI show that our stratified ensembling approach outperforms popular and state-of-the-art imbalance mitigation techniques. We also introduce a gold-standard evaluation protocol that uses unbiased test sets, and demonstrate that conventional hold-out evaluation from the same LMI-biased data can substantially mislead performance. Overall, our results highlight that the cause of imbalance is as important as the correction method.

Abstract: A central challenge in exploratory data analysis (EDA) is keeping track of what has already been examined in order to decide what to analyze next. In practice, analysts often run dozens of analyses while building an understanding of a dataset. However, most tools provide little support for maintaining an overview of this evolving process, instead exposing only a linear history of analysis steps. These tools show sequence, what came before, but not position, how a current analysis relates to the broader space of possible analyses. As a result, analysts must mentally reconstruct which parts of the space they have explored and where gaps remain, increasing the risk of redundant work or overlooked patterns. We present Ptolemy, a navigational interface that externalizes analysis history as a semantic map. Each analytic step is represented as a point positioned by embeddings derived from a structured description of its effective data view (e.g., columns, filters, transformations), allowing spatial distance to reflect analytic similarity. In a mixed-methods study comparing map, canvas, and tree representations, we find that maps improve global orientation and local comparison, while ordered layouts reduce decision cost. These findings surface a trade-off between orientation and actionability, and highlight design principles for supporting strategic exploration in EDA.