Telecom companies spend a lot to acquire new customers, making it crucial to retain these customers and ensure they continue using their plans. Churn refers to customers leaving their current service provider for another one. Once companies secure a long-term customer, they can profit significantly from them. Therefore, customer churn is a major concern for a company’s revenue and business growth, especially in the competitive telecommunications industry. By analyzing customer behavior and the services they use, companies can predict churn. This predictive capability allows them to minimize losses and enhance their business performance.

Pedestrian fatalities are a significant issue worldwide, particularly in the United States. These fatalities stem from various causes, including distracted driving, speeding, inadequate road maintenance, and poor visibility due to weather conditions. The risk also varies heavily by area and population density, with high-traffic and high-speed regions often prioritizing vehicle flow over pedestrian safety. The National Highway Traffic Safety Administration (NHSTA) collects extensive accident data involving pedestrians. Using this dataset, we aim to build a classification model to predict the likelihood of one or more fatalities in a traffic accident given key factors like road conditions, weather, time of day, speed and more. The goal of this research is to leverage predictive modeling to identify high-risk situations and develop intervention or fast reaction strategies. By combining NHTSA data with machine learning techniques, this study enhances our understanding of critical risk factors. Furthermore, it explores the potential for practical applications, such as improved road safety tools, safer urban planning, and real-time alert systems.

The prevalence of chronic diseases like diabetes, hypertension, and chronic respiratory conditions contributes significantly to global mortality rates. In addition, there is growing evidence that links mental health disorders to physical activity, body mass index (BMI), and chronic diseases, highlighting their importance in public health research. This study investigates the intricate relationship between chronic diseases and depression, employing innovative Machine Learning techniques to predict depression likelihood based on various health comorbidities. Results indicate that Naive Bayes consistently outperforms other models, highlighting its potential for accurate predictions. The trade-off between specificity and accuracy, however, highlights the necessity of balanced datasets in real-world applications.

Habitat loss, climate change, and disease have all contributed to the statewide decline in bat populations. Acoustic surveys are the least invasive method to determine bat community composition across the landscape. Acoustic mobile transects and stationary surveys were performed at three sites within the Altona Flat Rock, a globally rare sandstone pavement jack pine barrens in the fall of 2024. These sites were within the 2018 burn area, in an adjacent wildfire recovered (1957) jack pine (Pinus banksiana) stand, and in a mixed forest near the abandoned Million Dollar Dam in comparison to sites in agricultural and urban (SUNY Plattsburgh campus, City Marina) areas. We hypothesized that species richness would be higher in the 2018 burn site due to more open overstory and with close proximity to the wetland where insects should be more prevalent. Our results did not support this hypothesis with species richness being greatest at the Flat Rock forested dam (9). The Flat Rock burned site had a species richness of 6, and SUNY Plattsburgh campus had a species richness of 5. The Flat Rock unburned, or recovered site, City Marina and agricultural sites had the lowest species richness (4). In total, all nine native species were detected during our surveys, with the hoary and silver-haired bat being detected across all sites. Notably, we potentially detected two endangered species: the northern long-eared bat (Myotis septentrionalis) and the Indiana bat (Myotis sodalis), and a proposed endangered species, the tricolored bat (Perimyotis subflavus). We also found community similarity between the Flat Rock burned and Flat Rock forested dam, Flat Rock burned and City Marina, and Flat Rock burned and agricultural sites were the most similar (80%). Our results suggest that bats prefer more complex habitats like the Flat Rock forested dam. New York state is working to increase bat survey efforts by engaging citizen scientists. We plan to share our findings with local agencies and the New York State Natural Heritage Program to aid their efforts to map and conserve bat communities in unique habitats.

Freshwater crayfish are a group of macroinvertebrate omnivores that serve as a prey species for fish and birds. They are under threat with changing climates, and greater research on their distribution patterns across different habitats is needed to establish a necessary conservation strategy. Key factors which determine the presence of crayfish species are water velocity, water quality, over story, and substrate composition. Relatively limited research has successfully established a link between crayfish species and the water quality metrics. We predicted that fewer crayfish species and lesser abundance would be found in the lentic sites. In fall 2024, we placed 5 to 23 minnow and turtle traps in lotic (Little Ausable River) and lentic (Barracks Golf Course) sites for 24 hours. Crayfish captured were examined for parasites, in addition to being weighed and measured. Specimens were photographed, marked on the carapace with a permanent marker, and immediately released. Two species of crayfish were identified, one unique to each site, Faxonius immunis at the lentic ponds and Cambarus robustus at the lotic river. Additionally, greater crayfish abundance occurred at the lentic site. Water quality tests showed similar average pH (7.91, 8.02) and conductivity (198, 210) ranges between the lentic and lotic sites, respectively. Little Ausable River trapping site velocities ranged between 42.5 m3/s to 1.181 m3/s, which likely affected the crayfish population. Crayfish were on average smaller at the lentic versus the lotic sites, potentially as a function of predation, water quality, and available resources. By establishing a better understanding of the distribution of crayfish types of wetlands, more effective conservation strategies can be implemented for declining native species in North America.

Bird surveys are commonly actively performed using point count techniques; however, more recently, passive acoustics have become a complementary method to evaluate species richness and trends across microhabitats. Traditional methods like point counts may lead to biases in data records depending on the researcher's experience level. Autonomous recording devices (ARU), such as AudioMoths, are an affordable method for those studying avifauna to add to their toolbox. Confidence in rare species identification may increase by being able to leave AudioMoths out for a longer duration than feasible with traditional point counts and the ease with which artificial intelligence program workflows can attempt species confirmation. In the fall of 2024, we performed both passive (ARU) and active (point count) avifauna surveys across three microhabitats (e.g., wetland, forest, meadow) at Point au Roche State Park. Our active surveying method used binoculars and Merlin Bird ID to help identify species for three-10-minute intervals during each visit. Our passive surveying method used AudioMoths to record vocalizations deployed for seven days, recording during peak hours from 20:00-8:00. BirdNet-Analyzer was used to identify vocalizations from audio files. Species richness was greatest in the meadow (51) and wetland (47) and least in the forest (27). Merlin Bird ID recorded black-capped chickadee (Poecile atricapillus), Canada goose (Branta canadensis), northern cardinal (Cardinalis cardinalis), blue jay (Cyanocitta cristata), and downy woodpeckers (Picoides pubescens) across the sites. The American crow (Corvus brachyrhynchos) and barred owl (Strix varia) called abundantly at the wetland (43%) and the meadow (36%) sites, while the forest site was more limited. Avian community similarity was greatest between the wetland and meadow (64%) microhabitats. Other notable species observed are stripe-faced meadow katydid (Orchelimum concinnum), wood frog (Lithobates sylvaticus), and spring peeper (Pseudacris crucifer). We highlight the importance of conducting active and passive surveying across multiple habitats to more adequately capture rare species that might be difficult to capture with shorter survey durations.

Urbanization can affect wildlife populations in the form of road mortality, habitat degradation, and spatio-temporal behavioral shifts. Understanding how wildlife populations respond to these effects can help us manage and protect them. We studied painted turtles (Chrysemys picta) to determine how urbanization impacts their population structure, including abundance, age structure, and sex ratio. This project is in collaboration with the Ecological Research as Education Network (EREN) and guided by TurtlePop 2.0. We hypothesized that painted turtles would be more abundant in a rural setting over an urban setting due to better habitat quality. We predicted that there would be more female turtles in urban areas due to warmer temperatures driven by an urban environment and that there would be more juvenile turtles in the rural site due to less mesopredators. To evaluate urbanization, we monitored turtles along a gradient from rural at Deep Bay Pond located in Point Au Roche State Park to urban at the Barracks golf course pond complex in the City of Plattsburgh, NY. Turtles were captured using 10 baited crab hoop traps at each site over a 3-day capture session, visiting urban site 5 times and the rural site 4 times each in fall 2024. Captured turtles were measured in size, estimated in age and sex, notched with a unique three-letter code for mark recapture estimates, and released. We calculated and compared turtle abundance between sites and over time. In addition, we monitored water quality (e.g., pH, conductivity, pond and ambient temperatures), as well as pond size and basking site availability. The urban site had three more females than the rural site, as predicted and both sites had a higher proportion of males to females. Contrary to our prediction, the urban site had more juveniles than the rural site, with only one more juvenile at that site. Both sites had the same number of adults, but the urban site had more females to males with 8 females and 18 males. The rural site had 5 females and 21 males. The urban site had more females which supported our hypothesis, however, it also had one more juvenile than the rural site, refuting our hypothesis. All ponds had a slightly basic pH around 8. Temperature was slightly colder at the rural site, likely due to the larger size and connection to Lake Champlain. Conductivity was above 200 S/m in the rural site and urban ponds 5 and 6 while ponds 1, 2, and 4 were around 160 S/m. Basking sites did not seem to impact turtle abundance. The Barracks golf course may want to monitor any use of herbicides or fertilizers near the ponds while Point au Roche State Park may want to limit salting the road by the pond to reduce pollutants in the pond. Both urban and rural ponds are important habitats for painted turtles and other wildlife. Monitoring age and sex structure is important to assure that the population can successfully recruit and grow. Our study can be used to inform wetland management by highlighting the unique challenges that freshwater turtles face in both urban and rural settings.

Soundscapes are a compilation of sounds emanating from a landscape which creates an acoustic pattern across time and space. Wildlife behaviors are easily influenced by sound and can be reflected in periods of intense vocalization and silence. Urban landscapes are heavily influenced by anthrophony (human-produced sound) which can limit or enhance biophony (wildlife-produced sound). In fall 2024, the Wildlife Ecology and Management class arrayed five autonomous recording devices (AudioMoths) along an urbanization gradient (urban, urban interface, exurban golf course, rural riparia, and forested beaver pond). Audiomoths were programmed to record sound during peak hours of wildlife activity from 17:00-8:00 on two consecutive days during a warm spell. Data were downloaded and BirdNet-Analyzer was used to identify species using artificial intelligence (AI) algorithms associated with eBird. Output files were appended in R Studio and soundscape patterns were analyzed across sites. Biophony richness was highest at the exurban golf course (41), followed by urban (38), beaver pond (20), rural riparia (18), and urban interface (16). The most frequent callers at the urban site were American crow, dogs, squirrels, blue jays, Canada geese, and Carolina wren. At the urban interface, Canada geese, Carolina wren, and northern cardinals were highly vocal. At the exurban golf course, spring peeper, white-throated sparrows, barred owls, American crows, hairy woodpecker, and katydids were heard as compared to the rural riparia's eastern towhees, great-horned owls, and hairy woodpeckers. In contrast, anthrophonic call frequency was high and most similar at the urban (981) and and exurban golf course (891), intermediate at the urban interface (136), followed by the beaver pond (39), and rural riparia (33) with sounds including engines, humans, and power tools. At dawn the urban, exurban golf course, and beaver pond exhibited a high biophonic call frequency, while at dusk urbanized sites experienced extremely high calling rates compared to rural sites. Urban and exurban golf course communities were the most similar (51%), while the urban interface and the beaver pond were the least (22%). Soundscape ecology is a relatively new field that is growing in popularity due to the affordability of sensors and advances in AI. Our class learned that although these technologies have numerous benefits, there is still a need to perform supplemental surveys using traditional techniques or experts should be onboarded to help better identify misidentified calls.

Due to the volatility recently experienced in the United States Stock market, this study aims to explain the relationship between fundamental analysis and stock returns over a quarterly time frame to take advantage of the swings in the market. This study includes both micro (firm-specific) fundamental indicators and macro variables that help explain the economy's health as a whole. This study takes the thirty stocks currently in the Dow Jones Industrial Average (DJIA) as of February 2024 and analyzes their returns and underlying financials. It uses accounting and financial ratios along with measures of the overall economy to try and capture opportunities to generate financial returns in the Stock Market. This paper offers evidence based on company panel data analysis on a quarterly basis from quarter one of 2014 to quarter four of 2023. The results produced from the model indicate an increase in recession risk and federal funds rate generate negative stock returns. In contrast, an increase in the Price-earnings ratio and analyst recommendations generate positive stock returns. The above variables are significant at the 5% level with an R squared of 0.126.

This research paper examines the dynamics between economic development, financial stability, and crime rates across various countries. Rooted from the economic theory of crime and institutional anomie theory, this study uses a robust panel data analysis covering numerous countries to explore how economic metrics such as GNI per capita, foreign direct investment inflows, and financial institution strength relate to crime perceptions. The empirical analysis, using GLS, fixed effects, and random effects models along with a series of robustness checks, confirms significant relationships between these economic factors and crime levels. Preliminary results indicate that while economic fitness can occasionally spur increases in crime rates, possibly due to concentration of people, stronger financial institutions and a more stable government generally contribute to reductions in crime. This paper contributes to the ongoing discussion on crime economics by improving our understanding of the economic causes of crime and showing how changes in certain factors can reduce crime through economic strategies.