Competition in the wireless telecommunications industry is fierce. To maintain profitability, wireless carriers must control churn, which is the loss of subscribers who switch from one carrier to another.We explore techniques from statistical machine learning to predict churn and, based on these predictions, to determine what incentives should be offered to subscribers to improve retention and maximize profitability to the carrier. The techniques include logit regression, decision trees, neural networks, and boosting. Our experiments are based on a database of nearly 47,000 U.S. domestic subscribers and includes information about their usage, billing, credit, application, and complaint history. Our experiments show that under a wide variety of assumptions concerning the cost of intervention and the retention rate resulting from intervention, using predictive techniques to identify potential churners and offering incentives can yield significant savings to a carrier. We also show the importance of a data representation crafted by domain experts. Finally, we report on a real-world test of the techniques that validate our simulation experiments.

26Fishways designed for salmonids often restrict passage by non-salmonids and effective tools are 27 needed both to identify passage problems for non-target species and to inform remediation 28 planning. In this case study, we used migration histories from 2170 radio-tagged adult Pacific 29 lamprey (Entosphenus tridentatus) to identify locations of poor passage ("bottlenecks") at a 30 large, multi-fishway dam. Over ten years, 49% of tagged lamprey that entered fishways failed to 31 pass the dam. Models accounting for repeated attempts by individual lamprey indicated 32 successful passage strongly depended on attempted passage route. Success also varied with time 33 of fishway entry, water temperature, and lamprey body size. Most failed passage attempts 34 terminated in lower fishway segments, but extensive seasonal shifts in bottleneck locations were 35 detected. Ranking metrics helped prioritize bottlenecks and identified sites where structural or 36 operational modifications should improve lamprey passage. Our integration of spatially-37 intensive monitoring with novel analytical techniques was critical to understanding the complex 38 relationships between fishway features, environmental variation, and lamprey behavior. The 39 prioritization framework can be applied to a wide range of fish passage assessments. 40 41 42

Rising river temperatures in western North America have increased the energetic costs of migration and the risk of premature mortality in many Pacific salmon (Oncorhynchus spp.) populations. Predicting and managing risks for these populations requires data on acute and cumulative thermal exposure, the spatio-temporal distribution of adverse conditions, and the potentially mitigating effects of cool-water refuges. In this study, we paired radiotelemetry with archival temperature loggers to construct continuous, spatially-explicit thermal histories for 212 adult Chinook salmon (O. tshawytscha) and 200 adult steelhead (O. mykiss). The fish amassed ~500,000 temperature records (30-min intervals) while migrating through 470 kilometers of the Columbia and Snake rivers en route to spawning sites in Idaho, Oregon, and Washington. Spring- and most summer-run Chinook salmon migrated before river temperatures reached annual highs; their body temperatures closely matched ambient temperatures and most had thermal maxima in the lower Snake River. In contrast, many individual fall-run Chinook salmon and most steelhead had maxima near thermal tolerance limits (20–22 °C) in the lower Columbia River. High temperatures elicited extensive use of thermal refuges near tributary confluences, where body temperatures were ~2–10 °C cooler than the adjacent migration corridor. Many steelhead used refuges for weeks or more whereas salmon use was typically hours to days, reflecting differences in spawn timing. Almost no refuge use was detected in a ~260-km reach where a thermal migration barrier may more frequently develop in future warmer years. Within population, cumulative thermal exposure was strongly positively correlated (0.88 ≤ r ≤ 0.98) with migration duration and inconsistently associated (-0.28 ≤ r ≤ 0.09) with migration date. All four populations have likely experienced historically high mean and maximum temperatures in recent years. Expected responses include population-specific shifts in migration phenology, increased reliance on patchily-distributed thermal refuges, and natural selection favoring temperature-tolerant phenotypes.

An experimental fishway was used to examine adult Pacific lamprey, Lampetra tridentata (Gairdner), behaviour in a series of attraction and passage-performance tests. Among all experiments, lamprey oriented to the fishway floor and walls and were attracted to both ambient and concentrated flow. When confronted with highvelocity areas (vertical-slot and submerged-orifice weirs), many lamprey failed to pass upstream. However, lamprey were able to find and take advantage of low-velocity refuges when they were provided. Lamprey climbed shallow-and steep-angled ramps when attraction cues were sufficient and other passage routes were restricted. The combined results demonstrated the passage challenges that fishways designed and operated for salmonids present to non-salmonid species. They also highlight the importance of evaluating trade-offs between fishway attraction and passage efficiency. The experiments were integrated with tagging studies and development of lamprey-specific passage structures, a research combination that provided an effective template for fishway performance evaluations.K E Y W O R D S : burst swimming, Columbia River, fishway performance, migration barrier, rheotaxis, velocity refuge.

Optimization of fishways to pass multiple species is challenging because life history, swimming ability, and behavior often differ among species. For example, high fishway water velocities designed to attract adult Pacific salmon Oncorhynchus spp. at Columbia River dams inhibit fishway entrance and passage success of adult Pacific lampreys Lampetra tridentata, a species of conservation concern. We tested whether reduced water velocities (∼1.2 m/s, 0.15 m of head) at Bonneville Dam fishway openings improved entrance efficiency and other passage metrics for radio‐tagged Pacific lampreys compared with control velocities (>1.98 m/s, 0.46 m of head) and near‐zero (“standby”) velocities. Lamprey entrance efficiencies were significantly higher in the reduced‐velocity treatment (26–29%) than in the control (13–20%) or standby (5–9%) treatment. In some years, significantly more Pacific lampreys passed through fishway collection channels and transition pools and reached the fish ladder during reduced‐velocity treatment conditions, indicating that benefits extended beyond fishway entrances. However, overall passage efficiency at the dam was relatively unchanged, suggesting that additional passage bottlenecks for Pacific lampreys exist upstream from fishway entrances. The experiment demonstrated how operational changes can improve passage performance and how exploiting behavioral differences among species can improve multispecies management.