Connected cars are a rapidly growing segment of Internet of Things (IoT). While they already use cellular networks to support emergency response, in-car WiFi hotspots and infotainment, there is also a push towards updating their firmware over-the-air (FOTA). With millions of connected cars expected to be deployed over the next several years, and more importantly persist in the network for a long time, it is important to understand their behavior, usage patterns, and impact --- both in terms of their experience, as well as other users. Using one million connected cars on a production cellular network, we conduct network-scale measurements of over one billion radio connections to understand various aspects including their spatial and temporal connectivity patterns, the network conditions they face, use and handovers across various radio frequencies and mobility patterns. Our measurement study reveals that connected cars have distinct sets of characteristics, including those similar to regular smartphones (e.g. overall diurnal pattern), those similar to IoT devices (e.g. mostly short network sessions), but also some that belong to neither type (e.g. high mobility). These insights are invaluable in understanding and modeling connected cars in a cellular network and in designing strategies to manage their data demand.

User location is of critical importance to cellular network operators. It is often used for network capacity planning and to aid in the analysis of service and network diagnostics. However, existing localization techniques rely on user-provided information (e.g., Angle-of-Arrival), which are not available to the operator, and often require a significant effort to collect training data. Our main contribution is the design and evaluation of the Network-Based Localization (NBL) System for localizing a user in a 4G LTE network. The NBL System consists of 2 stages. In an offline stage, we develop RF coverage maps based on a large-scale crowd-sourced channel measurement campaign. Then, in an online stage, we present a localization algorithm to quickly match RF measurements (which are already collected as part of normal network operation) to coverage map locations. The system is more practical than related works, as it does not make any assumptions about user mobility, nor does it require expensive manual training measurements. Despite the realistic assumptions, our extensive evaluations in a national 4G LTE network show that the NBL System achieves a localization accuracy which is comparable to related works (i.e., a median accuracy of 5% of the cell's coverage region).

Understanding utilization of city roads is important for urban planners. In this paper, we show how to use handoff patterns from cellular phone networks to identify which routes people take through a city. Specifically, this paper makes three contributions. First, we show that cellular handoff patterns on a given route are stable across a range of conditions and propose a way to measure stability within and between routes using a variant of Earth Mover's Distance. Second, we present two accurate classification algorithms for matching cellular handoff patterns to routes: one requires test drives on the routes while the other uses signal strength data collected by high-resolution scanners. Finally, we present an application of our algorithms for measuring relative volumes of traffic on routes leading into and out of a specific city, and validate our methods using statistics published by a state transportation authority.

For users, recommendations can sometimes seem odd or counterintuitive. Visualizing recommendations can remove some of this mystery, showing how a recommendation is grouped with other choices. A drawing can also lead a user's eye to other options. Traditional 2D-embeddings of points can be used to create a basic layout, but these methods, by themselves, do not illustrate clusters and neighborhoods very well. In this paper, we propose the use of geographic maps to enhance the definition of clusters and neighborhoods, and consider the effectiveness of this approach in visualizing similarities and recommendations arising from TV shows.

A common task of recommender systems is to improve customer experience through personalized recommendations based on prior implicit feedback. These systems passively track different sorts of user behavior, such as purchase history, watching habits and browsing activity, in order to model user preferences. Unlike the much more extensively researched explicit feedback, we do not have any direct input from the users regarding their preferences. In particular, we lack substantial evidence on which products consumer dislike. In this work we identify unique properties of implicit feedback datasets. We propose treating the data as indication of positive and negative preference associated with vastly varying confidence levels. This leads to a factor model which is especially tailored for implicit feedback recommenders. We also suggest a scalable optimization procedure, which scales linearly with the data size. The algorithm is used successfully within a recommender system for television shows. It compares favorably with well tuned implementations of other known methods. In addition, we offer a novel way to give explanations to recommendations given by this factor model.

The collaborative filtering approach to recommender systems predicts user preferences for products or services by learning past user-item relationships. In this work, we propose novel algorithms for predicting user ratings of items by integrating complementary models that focus on patterns at different scales. At a local scale, we use a neighborhood-based technique that infers ratings from observed ratings by similar users or of similar items. Unlike previous local approaches, our method is based on a formal model that accounts for interactions within the neighborhood, leading to improved estimation quality. At a higher, regional, scale, we use SVD-like matrix factorization for recovering the major structural patterns in the user-item rating matrix. Unlike previous approaches that require imputations in order to fill in the unknown matrix entries, our new iterative algorithm avoids imputation. Because the models involve estimation of millions, or even billions, of parameters, shrinkage of estimated values to account for sampling variability proves crucial to prevent overfitting. Both the local and the regional approaches, and in particular their combination through a unifying model, compare favorably with other approaches and deliver substantially better results than the commercial Netflix Cinematch recommender system on a large publicly available data set.

Measuring distance or some other form of proximity between objects is a standard data mining tool. Connection subgraphs were recently proposed as a way to demonstrate proximity between nodes in networks. We propose a new way of measuring and extracting proximity in networks called “cycle-free effective conductance” (CFEC). Importantly, the measured proximity is accompanied with a proximity subgraph which allows assessing and understanding measured values. Our proximity calculation can handle more than two endpoints, directed edges, is statistically well behaved, and produces an effectiveness score for the computed subgraphs. We provide an efficient algorithm to measure and extract proximity. Also, we report experimental results and show examples for four large network datasets: a telecommunications calling graph, the IMDB actors graph, an academic coauthorship network, and a movie recommendation system.

Measuring distance or some other form of proximity between objects is a standard data mining tool. Connection subgraphs were recently proposed as a way to demonstrate proximity between nodes in networks. We propose a new way of measuring and extracting proximity in networks called "cycle free effective conductance"(CFEC). Our proximity measure can handle more than two endpoints, directed edges, is statistically well-behaved, and produces an effectiveness score for the computed subgraphs. We provide an efficien talgorithm. Also, we report experimental results and show examples for three large network data sets: a telecommunications calling graph, the IMDB actors graph, and an academic co-authorship network.

As large-scale deployments of spoken dialog systems in call centers become more common, a wealth of information is gathered about the call center business as well as the operation of these systems from their daily logs. This paper describes the "VoiceTone Daily News" data mining tool for analyzing this information and presenting it in a readily comprehensible and customizable form that is suitable for use by anyone from system designers to call center businesses. Relevant business and dialog features are extracted from the speech logs of caller-system interactions and tracked by trend analysis algorithms. We describe novel techniques for generating alerts on multiple data streams while avoiding redundant "knock-on" alerts. Some initial experiments with automated measures of dialog success are described as possible additional features to track. Features that move outside their expected bounds on a given day generate headlines as part of a website generated completely automatically from each day's logs. A "drill-down" facility allows headlines to be investigated all the way to viewing logs of individual interactions behind the headline and listening to the audio for individual turns.

A feature of data mining that distinguishes it from "classical" machine learning (ML) and statistical modeling (SM) is scale. The community seems to agree on this yet progress to this point has been limited. We present a methodology that addresses scale in a novel fashion that has the potential for revolutionizing the field. While the methodology applies most directly to flat (row by column) data sets we believe that it can be adapted to other representations. Our approach to the problem is not to scale up individual ML and SM methods. Rather we prefer to leverage the entire collection of existing methods by scaling down the data set. We call the method squashing. Our method demonstrably outperforms random sampling and a theoretical argument suggests how and why it works well. Squashing consists of three modular steps: grouping, momentizing, and generating (GMG). These three steps describe the squashing pipeline whereby the original (very large data set) is sectioned off into mutual...