Heatmap is one of the effective representations for time-varying data visualization. It may require large display spaces when an input dataset contains large number of data items or time steps. We may often want mechanisms to interactively filter non-important data items or time steps, so that we can form appropriate sizes of heatmaps and focus on important data items or time steps. This paper presents a heatmap-based time-varying data visualization technique featuring an interactive mechanism to display meaningful data items and time steps. This technique firstly calculates distances between arbitrary pairs of data items, and constructs a dendrogram consisting the data items. It then generates clusters of the data items and displays the data items belonging to the specified sizes of clusters in the heatmap, so that we can focus on groups of similar or correlated data items. It applies a similar mechanism to a set of time steps so that we can remove outlier time steps from the heatmap. Our implementation features two scatterplots, which represent distribution of data items and time steps respectively, and slider widgets to interactively adjust the thresholds of the clustering process. We can intuitively understand how clusters of data items or time steps are constructed, by looking at the scatterplots while operating the sliders.

Regression analysis has been widely applied to various academic and industrial fields. Applications of regression analysis include medical problems such as health estimation, environmental problems such as disaster prediction and energy consumption estimation, and business/economic analytics. Here, accuracy and quality of regression analysis strongly depend on relevancy between input explanatory variables and actual objective functions. It often happens that several explanatory variables are well correlated with objective functions while others do not well correlated, and therefore accuracy of regression analysis may improve by removing unnecessary explanatory variables. This paper presents a scatterplot-based regression analysis tool. This tool visualizes the distribution of errors between actual and estimated values of objective functions, and provides user interfaces to explore the relationships between explanatory variables and the errors. This paper introduces examples of the visualization results using the presented tool with actual and estimated revenues at a store.

Model interpretability has been recognized to play a key role in practical data mining. Interpretable models provide significant insights on data and model behaviors and may convince end-users to employ certain models. In return for these advantages, however, there is generally a sacrifice in accuracy, i.e., flexibility of model representation (e.g., linear, rule-based, etc.) and model complexity needs to be restricted in order for users to be able to understand the results. This paper proposes oblique treed sparse additive models (OT-SpAMs). Our main focus is on developing a model which sacrifices a certain degree of interpretability for accuracy but achieves entirely sufficient accuracy with such fully non-linear models as kernel support vector machines (SVMs). OT-SpAMs are instances of region-specific predictive models. They divide feature spaces into regions with sparse oblique tree splitting and assign local sparse additive experts to individual regions. In order to maintain OT-SpAM interpretability, we have to keep the overall model structure simple, and this produces simultaneous model selection issues for sparse oblique region structures and sparse local experts. We address this problem by extending factorized asymptotic Bayesian inference. We demonstrate, on simulation, benchmark, and real world datasets that, in terms of accuracy, OT-SpAMs outperform state-of-the-art interpretable models and perform competitively with kernel SVMs, while still providing results that are highly understandable.

Most time-varying data in our daily life is multi-variate. Moreover, most of such time-varying data contains both numeric and categorical values. It is often meaningful to visualize both of them as they are often correlated. We aim to visualize every value in such time-varying data in a single display space so that we can discover interesting relationships among the values of the time-varying data. This paper presents a heat map-based time-varying data visualization technique which displays both numeric and categorical values in a single display space. The technique assigns time to the horizontal axis of the display space, and vertically arranges the series of colored belts corresponding to the time-sequence values. It generates one belt for a numeric value, and multiple belts for a categorical value. It clusters the belts according to the similarity of color sequences, and re-arranges the belts based on the clustering result. This paper shows an example of the visualization result applying a time-varying multi-variate marketing dataset.