2020 Student Projects

How much should I charge for my Airbnb? Pricing AirBnB using descriptive and predictive analytics

Author(s): Lei Wang (Computer Information Systems)

Faculty Mentor(s): Sonya Zhang

Abstract:Our research study intends to identify the key factors affecting the listing price of AirBnB using descriptive and predictive analytics. As Airbnb rises as a new form of tourist accommodation, many travelers enjoy staying in Airbnb housing due to their low costs, convenient locations, household amenities, and personal services. However, the owner of the property, or the host, may not have the best knowledge at setting the right price for their property. A rough scan of local neighborhood prices may not be enough or accurate as there are many differences between properties. Our study is a multi-step process: First, we extracted and cleaned the Airbnb data set which contains over 28,000 listings in Los Angeles County. Second, we performed descriptive analytics to summarize main characteristics using visualization. Third, we used several different machine learning algorithms to develop the predictive model for the listing price based upon a collection of features such as type of property, number of rooms, location, reviews, etc. Outside resources, such as median household incomes by zip code, were also included in our models..

Summary of Research Findings:We developed multiple regression and classification models to help Airbnb hosts better predict the pricing of their short term rental properties..

Key Words: Data Analysis, Machine Learning, Feature Selection, Airbnb

Do Mentorship Programs and Supportive Environment Encourage Women to Enter Leadership Positions?

Author(s): Sheena Devadoss (Management and Human Resources)

Faculty Mentor(s): Chantal van Esch

Abstract:My hypothesis is that if aspiring women are given proper guidance and supportive environment, then there is a higher chance that they will aspire to to enter managerial ranks. This hypothesis is based on the process of how men move up from support roles to managerial roles and then to executive ranks..

Summary of Research Findings:Based on my research, women make up a small portion of the executive ranks in the corporate world. Of the 1,000 companies “that make up the 2019 Fortune 500 list, 33 have women CEOs” (Connley, 2019). Over the last 30 years, there has been some improvement. However, the increase of women in executive ranks has been minimal. As mentioned in the Wall Street Journal article, “if companies in the U.S. continue to make the same, tiny gains in the numbers of women they promote and hire into management every year, it will be another 30 years before the gap between first-level male and female managers closes” (Fuhrmans, 2019)..

Key Words: Women, Leadership, Mentorship, Management

Should Business Faculty Encourage Their Undergraduate Students to Participate in Research?

Author(s): Cailin Kuchenbecker (Masters of Business Administration), Guillermo Marquez (Marketing), Mitchell Pickering (Computer Information Systems)

Faculty Mentor(s): Jae Min Jung

Abstract: Psychology literature shows undergraduate students who passively participate in research (e.g. survey-takers) find the experience valuable and have a positive impact on their education. Although the value of research participation (RP) has been established in psychology literature, similar research in marketing/business is limited. The lack of acceptance by business faculty of the positive outcomes from RP creates potentially missed opportunities for learning. Thus, this study identifies various factors that might influence the effectiveness of passive RP, which include: perceived value of RP in business research, helpfulness of RP in learning marketing, student’s attitudes towards participation, and student’s intentions to participate in future studies. Data was collected from students enrolled in marketing courses. Linear regression and t-tests were used for analysis. Results indicate that students’ motivation to learn marketing and higher levels of sincerity and interest were associated with student’s reporting that RP was more valuable and more helpful in learning marketing. These students also held more positive attitudes towards the research and are more willing to participate in future research. Furthermore, business (vs. non-business) majors found RP to be more helpful in learning marketing and were more inclined to participate in research again. Business majors also found the research to be more interesting than non-majors, suggesting participant’s interest is an important factor determining students’ perception of helpfulness of research in learning and their intention to partake in future studies. This research provides insights for business literature by providing evidence of the effectiveness of experiential learning in business education..

Summary of Research Findings: H1, H2, and H4 were tested with independent samples t-tests. Consistent with H1, results showed that participants with both types of motivations (intrinsic and extrinsic), compared with those with only extrinsic motivation, perceived RP to be more effective (p’s < .001) in terms of all four dependent variables. Results of four t-tests revealed that consistent with H2, extra credit was more effective (vs. required credit) for inducing higher intention to participate in the future (p < .001); however, unlike hypothesized, it was not not more effective than required credits (p’s > .10) in inducing (a) value of RP, (b) helpfulness in learning marketing concepts, nor (c) attitudes toward RP. Thus, H2 was partially supported. To test H3, we regressed each of the four dependent variables on sincerity, interest, and major (business vs. non-business), with the latter two as control variables. Results confirm the positive impact of sincerity on RP effectiveness (p’s < .05), consistent with H3. Results of t-tests show that business (vs. non-business) majors perceived RP to be more helpful in learning marketing (p < .001) and indicated stronger intention to participate in the future (p < .001), but majors had no impact on the value of RP nor on attitudes toward RP. Thus, H4 was partially supported. Results of a t-test show that business majors (vs. non-business) majors had stronger interest in RP (p’s = .06; H5), which in turn is positively related to RP effectiveness (p’s < .001; H6). H4, H5, and H6 together suggests that interest in RP may mediate the impact of majors on RP effectiveness. Indeed a procedure by Baron and Kenny (1986) reveals that it does partially mediate the impact of majors on RP effectiveness. .

Key Words: Value of Passive Research Participation, Business Pedagogy, Undergraduate Education, Marketing Pedagogy

Use of Spokespersons in Advertising Gendered and Neutral Products

Author(s): Cailin Kuchenbecker (Masters of Business Administration)

Faculty Mentor(s): Jae Min Jung

Abstract: Spokespersons have long been used in marketing to promote products, with men being used more frequently in product advertisements. Interestingly enough, research has shown that most products are perceived as either feminine (e.g. hand lotion), masculine (e.g. trucks), or neutral (e.g. toothpaste). Additionally, research from the 1980’s found that matching a spokesperson's gender to the gender of the product in advertisements generally creates more positive perceptions from the audience than if the spokesperson’s gender does not match that of the product. While past literature has focused on gendered products, it does not fully address the role neutral products play in spokesperson advertising. In addition, with gender roles rapidly changing in the United States, it is important to reevaluate past findings studied over 30 years ago. Therefore, this study investigates the durability of past research and explores how best to promote neutral products using spokespersons to attract consumers. An experiment was conducted on millennial college students and t-tests were used for analysis. This study provides insights on how to approach and utilize spokespersons in advertising gendered/neutral products and adds to the literature on gender perceptions in society..

Summary of Research Findings: Hypothesis 1 states that matching a spokesperson’s gender to the gender of the product in advertisements will have no effect on individual’s (H1a) attitudes, (H1b) interest levels, or (H1c) purchase intentions of said product. The gendered product used was beer [M = 5.6 on masculinity scale (1 = Extremely feminine - 7 = Extremely masculine)]. To test H1a, H1b, and H1c, several t-tests were used. All parts of hypothesis 1 were supported, showing that matching/mismatching spokespersons gender and product gender do not effect attitude (H1a; M = 4.15 vs. M = 4.28, p > 0.05), interest levels (H1b; M = 3.01 vs. M = 3.18, p > 0.05), or purchase intentions (H1c; M = 3.25 vs. M = 3.51, p > 0.05) of the product. H2a and H2b state that female (vs. male) spokespersons promoting a neutral product will alter the product to be perceived as more feminine (vs. masculine). The neutral product used was bath soap [M = 3.7 on masculinity scale (1 = Extremely feminine - 7 = Extremely masculine)]. To test H2a and H2b, I used several t-tests. H2a was supported, which showed that female spokespersons promoting a neutral product will alter the product to be perceived as more feminine (M = 3.52, p < .001). On the other hand, H2b was not supported (M = 3.93, p > 0.05)..

Key Words: Gendered products, Neutral products, Spokespersons, Advertising

Professors, Disabilities, and Other Social Identities - An Intersectional Look at Higher Education

Author(s): Rosa Ramirez (Marketing Management)

Faculty Mentor(s): Chantal van Esch & Shayda Kafai

Abstract: There is extensive research on students with disabilities but there is a research gap when it comes to faculty with disabilities. The purpose of this study is to identify if faculty members who self-identify as having a disability (visible or invisible) are receiving the proper accommodations and accessibility in the workplace. This qualitative study centers on the participants intersectionalities (social identities) to dive deep into the impact of employee development in academia. This research seeks to answer what type of direct and indirect discrimination exist in their world of higher education. Some of the questions asked in the interviews are: Have you faced additional difficulties as a faculty member because of your disability? What are other social identities that have had an impact on your career, do you believe that they have intersected with your disability in terms of impact?There is extensive research on students with disabilities but there is a research gap when it comes to faculty with disabilities. The purpose of this study is to identify if faculty members who self-identify as having a disability (visible or invisible) are receiving the proper accommodations and accessibility in the workplace. This qualitative study centers on the participants intersectionalities (social identities) to dive deep into the impact of employee development in academia. This research seeks to answer what type of direct and indirect discrimination exist in their world of higher education..

Summary of Research Findings: Preliminary results demonstrate that faculty members with disabilities experienced three key challenges related to accessibility and support on campus: (1) stress and anxiety about disclosing their disability status; (2) obstacles in obtaining accessibility in their schedule and transportation on campus; and (3) fear of their RTP (Retention, Tenure and Promotion) application being denied due to their disability and being subjected to direct and/or indirect discrimination..

Key Words: accessibility, accommodations, disabilities study, faculty

Disney World Attractions: An Analysis of Wait Times and Sentiment

Author(s): Kevin Lee (Computer Information Systems, Technology and Operations Management)

Faculty Mentor(s): Rita Kumar

Abstract: As the availability of data from an expanding array of data streams grows exponentially, organizations and analysts are now increasingly reliant on data analytics techniques to generate insights for informed decision-making and to spearhead process improvements and innovations. The amusement park industry is no different: timely data analysis can be indicative of current conditions, influencing the daily operations and allocation of resources within Disney World and its selection of attractions and rides. Indeed, previous research applications of data analysis have led to mobile apps such as Lines, which utilize information such as historical wait times and live location data from app users to predict wait times with astonishing accuracy. The implications of such tools are extensive. To start, tourists can more easily plan trips by strategically minimizing wait times; Disney itself can utilize this data to manage demand and proactively plan services. In this study, we analyze a dataset of wait times for a selection of rides from 2012-2019. Specifically, we employ data visualization and several analytic techniques to understand the overall trend and the various factors impacting Disney World’s wait times. In addition, we combine these factors with time series forecasting techniques to predict wait times at a daily level..

Summary of Research Findings: Disney posts their own projected wait time for each ride, and in our data we notate this as "posted wait times". Also in our data is "actual wait times", which were the actual times obtained in person and gathered by touringplans.com. We pre-processed the raw data given, aggregated it at a daily level, then visualized it in an interactive Tableau dashboard. Through an interactive dashboard, we were able to see interesting trends and patterns for various scenarios. In general, there was a high disparity between actual and posted wait times for most rides, with average posted wait times consistently higher than actual wait times. We also performed seasonal decomposition by month, thereby obtaining seasonal factor scores for both posted and average wait times to determine the extent to which seasonality affected the wait times. Both are notably impacted depending on month and season, to varying degrees. In addition, we obtained a sentiment score on our selection of rides by applying the TextBlob and VADER language processing algorithms on tweets collected within the time frame of the respective ride. These algorithms rank tweets from a -1 to 1 scale, indicating negative and positive sentiment, respectively. From our result, we have found that tweets regarding our selection of rides have mostly been on the positive side, with Spaceship Earth scoring the highest..

Key Words: Data Visualization, Tableau, Sentiment, Forecast

Does the Perception Meet Reality? An Exploration of the Existing Stereotypes of Accounting Students

Author(s): Alison Arcos (Accounting)

Faculty Mentor(s): Preeti Wadhwa

Abstract: Accountants have been stereotyped as dull and uninteresting, consequently, the occupation of accounting has been portrayed in a lackluster way in popular media. These negative stereotypes could influence young adults who want to pursue accounting as their field of study, thereby making them reluctant to proceed with their choice. This research aims to examine the personality traits and other characteristics (including campus involvement) of current undergraduate accounting students, and then compare these personality profiles with the perceptions/stereotypes that non-accounting majors (freshmen and sophomores) in the business field have of accounting students. It further examines if these stereotypes of accounting students held by non-accounting students (seniors) change upon exposure to accounting undergraduates as they progress through their business degree. Data was collected via online Qualtrics surveys from a representative sample of undergraduates from CBA. Students took these surveys for extra credit. Specific hypotheses were tested. Preliminary results indicate that the negative stereotypes of accounting students do exist. Interestingly, the accounting majors themselves tend to use more adjectives with negative stereotypical connotations to describe themselves compared to non-accounting majors’ descriptions of accounting students. This indicates that they may have internalized these stereotypes. Additionally, having more friends pursuing accounting results in more positive stereotypes of accounting students as held by non-accounting majors. More data is being collected to get a deeper insight on our specific hypotheses. In terms of implications, hopefully, these findings will encourage prospective students to develop more positive images of accounting majors and professionals. Other implications are also discussed..

Summary of Research Findings: I was able to determine the Big 5 from various majors within the College of Business and compare them with the Big 5 of accounting majors. Furthermore, I was able to analyze data based on open ended free response questions regarding non accounting majors opinions of accounting majors and vice versa..

Key Words: creative thinking, accounting, personalities, stereotypes

Which restaurant to choose? - Mining the topics and sentiments of Yelp restaurant reviews

Author(s): Linda Ly (Computer Information Systems), Norman Mach (Computer Information Systems), Christian Amaya (Computer Information Systems)

Faculty Mentor(s): Sonya (Xuesong) Zhang

Abstract: Our research study intends to develop a recommender system for restaurants based on the topic and sentiment of online reviews using text mining. Many products or services now have a large number of online reviews, which makes it difficult for consumers to decide which reviews to pay attention to, or for businesses to make decisions which areas to improve on. Our prior studies (Zhang et al., 2018; Salehan et al., 2017) developed a content filtering recommender system using eight restaurant review factors from the literature review, and a human coded Yelp dataset. This study extends the recommender system by automating topics extraction and sentiment analysis. We also used a more updated Yelp dataset, which, after cleaning, contains 11,609 restaurant reviews. We used a Natural Language Processing method called Latent Dirichlet Allocation (LDA) to extract topics from the reviews and compare them with the literature review findings. Then we used a Python library called TextBlob to determine the sentiment of each review at sentence-level, which can be used to find sentiment for each review, review topic, and more customizable categories..

Summary of Research Findings: Research of topic extraction and sentiment analysis for Yelp restaurant reviews with the intention of building a better, more useful recommendation system for Yelp..

Key Words: Topic Modeling, Sentiment Analysis, Data Analytics, Text Mining

An Exploration of the Modifying Role of State Ethnocentrism on State-of-Origin Effects

Author(s): Jarrod Griffin (Computer Information Systems), Jillian Muñoz (Psychology), Cailin Kuchenbecker (Masters of Business Administration), Stephanie Muñoz (Finance, Real Estate, and Law)

Faculty Mentor(s): Jae Min Jung

Abstract: Currently, many U.S. state governments have state-labeled logo programs (eg., CA Grown, Ohio Proud, Pride of Dakota), which allow local businesses to affix origin information to the products they market within and outside their states (e.g. CA Grown sticker). With numerous social movements becoming more popular (e.g., Farm-To-Table and Community Supported Agriculture), the public is becoming increasingly aware of the origins and sustainability of their food. Consumers have responded positively to state-labeled logo programs, boosting local food sales significantly and drawing attention from industry leaders and academics alike. Thus, comprehensive research is needed to understand consumer attitudes and motivations for buying local products. To this end, we first systematically search for articles on the topic from agricultural economics and marketing literatures, synthesized and integrated past research, and developed a framework that will facilitate future research. Further, we investigate consumers’ attitudes and purchase intentions of the products made in own (vs. other) states and assess factors that could influence consumers' attitudes and purchase intentions. Data was collected from 528 students from two different state universities located within the United States. Results indicate that consumers have more favorable attitudes towards and greater purchase intentions for the products made in their own (vs. other) state and that such SOO effects were further moderated by the level of state ethnocentrism held by residents of the state. This research provides insights into government agencies and marketing literature by extending country-of-origin research and investigating state-of-origin effects in a novel way..

Summary of Research Findings: A systematic search of four databases — ABI/Inform, Agricola, Business Source Premier, and Web of Science — was conducted using a combination of keywords including province, state, region, consumer, bias, locally produced, preferences, branding, and grown. This process has led to the identification of forty-seven peer-reviewed articles focused primarily on consumer attitudes towards locally produced products. Content analysis indicates that the vast majority of the articles (91.49%) were about the food products while only three articles (6.38%) were about non-agricultural, consumer goods. The articles studied consumers not only in the United States (42.55% of articles) but also overseas (57.45%) across Europe (38.30%), Asia (10.64%), South America (4.26%), Canada (2.13%) and Australia (2.13%). The findings show that consumers around the world are willing to pay a premium for locally produced goods. In addition, to address consumer attitudes and purchase intentions of the products made in their own (vs. other) states, we conducted a 2 (SOO: North Dakota vs. Ohio) x 2 (Argument Strengths: Strong vs. Weak) between-subjects experiment using data collected from residents from the two states. Thus, the stimulus used was four different types of advertisement copies for a mock cereal brand that vary only on state-of-origin (ND vs. OH) and on argument strength (Strong vs. Weak). To test H1 (consumers’ tendency to support locally made goods), we created localness (local vs. non-local), a new variable, by grouping ND or OH residents who were exposed to a mock cereal brand made in their own state as “local” and those residents who were exposed to a mock cereal brand made in another state as “non-local.” The two groups were compared with their attitudes and purchase intentions. Results show that participants expressed more favorable attitudes (M = 4.82 vs. M = 4.53; F = 3.89; p = .05) and stronger intentions to purchase the brands (M = 3.77 vs. M = 3.21; F = 10.02; p = .01) when the products originated from their own state rather than the other states. To test H2 (moderating impact of state ethnocentrism), we ran a 2 (Localness: local vs. non-local) x 2 (Residency: ND vs. OH) MANOVA on attitudes and intentions with state ethnocentrism, vertical individualism, and vertical collectivism included as covariates. Results show a significant multivariate main effect of vertical collectivism (p < .05) and state residency (p = .05), More importantly, there was a significant multivariable interactive effect of localness x state residency (p = .01). We also confirmed that ND residents are more ethnocentric than OH residents (M = 3.26 vs. M = 2.74; t = 3.31, p = .001). Subsequently planned comparison was conducted separately for ND (high ethnocentric state) and OH (low ethnocentric state). Results showed that the state of origin effects manifested in H1 was amplified among ND consumers (Attitudes: M = 4.99 vs. M = 4.38, t = 2.30, p < .05; intentions: M = 4.37 vs. M = 3.24; t = 3.76, p < .001), whereas the SOO effects were attenuated among OH consumers (Attitudes: M = 4.70 vs. M = 4.80, t = -.47, p > .60; intentions: M = 3.35 vs. M = 3.56; t = -.82, p > .40), consistent with H2. .

Key Words: Consumer Preference, Local Labels, Locally Produced, State Labels

PCV MURCOR REAL ESTATE ANALYSIS

Author(s): Patrick Ogaz (Computer Information Systems), Feiyu Han (Finance, Real Estate, and Law), Mark Gordon (Technology and Operations Management), Sho Ishimaru (Finance, Real Estate, and Law), Annette Bedard (Technology and Operations Management)

Faculty Mentor(s): Rita Kumar & Anthony Orlando

Abstract: Playing a crucial role in the process of housing transactions, real estate appraisers provide an objective and unbiased estimate of the value of the property for the lenders. Appraisal Management Company (AMC), on the other hand, is the medium that connects lenders and appraisers, while creating efficiency and oversight between the two parties. PCV Murcor is one of the nation’s leading Appraisal Management Companies. During the Fall semester of 2019, our team worked closely with the company’s management and analyzed over 100,000 order records of the company from 2017-2019. We performed a descriptive analytics exercise to understand the overall trend and various factors impacting PCV Murcor’s appraisal quality. In addition, by utilizing publicly available data such as Zillow's Assessor and Real Estate Database, we identified the company’s scale and market position in each state..

Summary of Research Findings: Utilizing PCV data, Zillow's Assessor and Real Estate Database (ZTRAX), and Tableau we were able to analyze PCV to better understand how it fits into its market. We first shifted our emphasis on how appraisal fees compare to the actual appraisal price. This was due to the insights that PCV gave us about the appraisers being contract employees, meaning appraisal fees were established by the appraiser. We examined and compared the national median home prices to the average fees that PCV had recorded in each state. This provides PCV with information about what areas not only had high fee costs but also states with expensive median prices. Using Zillow’s data we found states like California, Colorado, and Washington having the highest median home prices suggesting higher fees in these states would look normal. It also was an indicator of the quality of PCV’s properties received from banks. A higher than national median would suggest a good quality while low would establish lower quality properties. This gave PCV a holistic look on their market position compared to its competitors. When evaluating median values by state, we evaluated the change in median price over the years 2017-2019 to see if there were any states or regions that were specific outliers. Zillow data suggests housing price changes in the midwest region shows the most increase, with 30.2% for Idaho, 26.18% for Utah, and 25.18% for Nevada. When considering median sales price we wondered if an increase in median prices meant we could observe higher profits for PCV. Due to the revenue structure of an AMC, we explored the change in the number of homes sold in those states using publicly available data. What we found is in states like Utah (near salt lake city), sales volume decreased even as the prices (median) increased. Suggesting profit for a company like PCV, which relies on volume, could be affected as there is a decrease in the number of flat fees collected. Through this analysis, we were able to better understand where PCV fits into its market. By using publicly available data, we were able to establish areas of success or improvement to help PCV better navigate success..

Key Words: Real estate, Appraisal, PCV Murcor, Zillow