Sneha Verma
The potential of data to support society is unmatched and my interests and education have taught me that in a globalizing community, data is a necessary tool to create national and international cooperation and awareness. My name is Sneha Verma and I am a Data Science and Economics junior at Luther College. I believe that data can and should be used to solve social problems, create informed policies, promote transparency and social awareness, and innovate.
I am from India and have lived in six countries over a period of 17 years. I have been introduced to a multitude of cultures, beliefs, and academic curriculums. An important lesson that I have learned from the experience of living in different countries and meeting people with varying backgrounds is that diversity is not to be feared, it is to be revered and accepted; differences are not meant to demean, rather be learned from and appreciated. In a world that is connected through technology but divided because of beliefs, I believe that data can be used in all fields and industries to bridge the gap between ignorance and awareness, between fear and reverence. My interest in economics fosters my passion for using data to solve existing problems, to create policies that help organizations and governments create data-backed policies, efficient and necessary products and services, and create a transparent environment between people of authority and the public.
At Luther College, I am exploring the intersection between economics and data science. I have completed various projects and participated in multiple extra-curricular activities to explore the two fields.
Credentials:
I completed 11th and 12th grades in this school. During these two years, I worked as the Assistant Manager for a theatre production which gave me leadership experience. I graduated as the salutatorian of my grade.
I am supporting strategic business plans by analyzing data, maintaining queries, and creating documentations.
... I utilize SQL, Excel, Power BI, Power Query, and other tools and languages to accomplish these tasks and develop reports, dashboards, and products. I employ the aforementioned tools and Jira to create and deploy development products, test results and finalize proper implementation alongside the business owners and stakeholders to complete the software development and project lifecycle. I identify and pursue career and personal growth opportunities through company events, volunteer and community service actions, and networking.
I worked with the Marketing Operations Director at Luther College to conduct data analyses, audit the website and manage the CRM system.
... My responsibilities included exploring business opportunities and problems and finding solutions and insights with data. I conducted data entry and analyses that inform decision-making about enrollments and operations. Further, I helped maintain, Slate, the customer relationship management system and supported good data hygiene, efficient email segmentation, and accurate analysis. I was engaged in multiple projects involving analyzing digital and email communication data. Using data collected from 5+ years, I collected emails sent to prospective students to analyze various aspects to understand how to increase open and click rates. With the results obtained, I liaised with the Admissions team to inform their decision-making and operations of the results of my research.
I worked with the Performance Team that oversees Thomson Reuters' legal directory products. My responsibilities as a Digital Marketing and Analytics Intern included automating tasks, analyzing data to create machine learning web products, and analyzing and visualizing web analytical data to inform business decisions.
... I have automated data analysis tasks reducing time spent on debugging code by 100%. This has led to the efficient creation of 28 blogs on legal information. Further, I worked on creating a user-friendly, machine learning model that will produce relevant legal information based on various features. For this project, I explored various machine learning algorithms, including classification and neural networks. I built new analytic capabilities with Python and R by working with APIs and analyzing key metrics in web analytical data to inform decision-making with Python and Power BI.
As the Web and Data Analytics Intern, I was responsible for working on the Diversity Council's website, creating efficient process management systems, and working with an external organization to encode cultural competency questions.
... I worked on administrative projects such as updating documents and annual reports for the 2021 year. Further, I worked with Microsoft products (SharePoint, Power Automate, Forms, etc.) to create a database that stored cultural, religious, LGBTQIA+, and community resources. This database is linked to the Diversity Council website and is accessible by the public. A large and influential part of my internship was working with Cradle 2 Career, a community-wide and community-owned initiative that is dedicated to educational equity. I worked with the Director of Community Engagement and a Winona State University professor, along with another intern, to create a cultural competency test that can be used to assess the cultural competency of educational and youth support systems.
As a media relations assistant, I wrote over 10 press releases about Luther College in Associated Press (AP) style. I was also responsible for researching avenues for press releases and media pitches and communicating with the Luther community to gather information. I tracked media placements of Luther College and managed the department's email account.
... I wrote press releases highlighting the achievements of Luther students, staff, and faculty members by communicating with members of the institution to receive reliable information. As a student worker, I was responsible to find and research stories and communicate with people to spread awareness.
I looked over 20 resumes to help students of all grade levels apply for internships and have had numerous meetings to do the same. The Career Center Instagram and Facebook accounts reached over 300% more engagement and people are interacting with the Career Center more through social media. I created picture posts and stories for the Career Center to promote the department's services.
... As a Career Peer Advisor, I supported and encouraged fellow students throughout their career process from developing self-awareness to internship and job searches, to crafting resumes and cover letters and preparing for interviews. I also assisted the office staff with projects that involved keeping records of various events and workshops. Further, I designed and helped implement student work development and contributed to the diversity, equity, and inclusion of the department. I also handled the website and social media accounts along with acting as a liaison to affinity groups.
As the office assistant, I helped in conducting pedagogical research and organise events, workshops, and meetings for faculty development. Further, I completed administrative tasks around the office, such as updating information on speaker events, workshop attendance, and one-on-one conferences with the CELT director.
... I also collected data to understand and assist in the improvement of teaching methods across campus. I acted as a liaison to other departments to receive data and conduct researches to inform and better teaching practices at Luther. Data collected include those from surveys and research. I worked on creating content for the CELT department, such as efforts to aid professors to adjust to the needs of the pandemic, creating content for and editing the CELT website by editing recorded videos of professors sharing their teaching techniques. I created multiple CELT videos, and I analyzed student surveys to create data spreadsheets to inform professors about the effectiveness of their teaching techniques leading to an increase in the traffic to the CELT website.
The research, with Dr. Samuel Bird, studies economic decision-making in the context of agriculture in sub-Saharan Africa, one of the most important economic sectors and regions for economic development.
... In particular, the project studies whether the decision of harvesting method differs between men and women within a household. This project adds to the research of decision-making models in the agricultural fields while investigating gender norms under an economic view. For this research, I collected the data and cleaned it with the Stata 16 software. Further, I conducted exploratory data analysis through Stata do-files. I created data visualizations and tables to demonstrate the variance and skewness of the variables and to display the final analysis results obtained through multiple linear regression. The results are documented on a data appendix created by the Stata Markdown language. This research has been presented at the AAEA conference with a theory paper written by myself and Dr. Samuel Bird.
I was accountable for training new hires, creating training programs, and assisting technicians through problem-solving. I worked closely with the technology help desk managers and technicians.
... I took part in the hiring process by reviewing applications, conducting interviews, and recommending changes to the managers. Projects I have participated in include scheduling fun activities to boost morale, creating a semester-long game to create a fun environment during work, and managing the department's social media handles. I created content for the Technology Help Desk Instagram and Twitter accounts, such as Tech Tips and Techie Tuesdays, to increase student engagement with the department. Post engagement increased by 20% and followers increased by 10%.
I worked in the Trauma Centre and Safdarjung Hospital Dharamshala assisting patient outtakes and handling administrative tasks with my fellow interns.
... Further tasks that I conducted were distributing food to the sick and destitute in the Safdarjung Dharamshala and visiting patients in the Trauma Centre and providing technical and administrative assistance to the social workers. I assisted in inpatient care and data and budget management. At the end of the internship, I created a report describing the successes of the organization by listing the number of patients that were assisted, that are living in the housing facility, that are provided food in the Dharamshala, and the donations that were received.
This research focusses on the probabilities fof students from a certain race, gender, or financial status. The data was collected from the websites of individual colleges. This was my senior research project in Economics at Luther College.
For further details, read more or view the project!
... Through this project, I wanted to analyze the effects of race, gender, and financial status on the chances of graduating from a Midwestern Liberal Arts college. A further consideration of this proejct is to combine these factors into one model to understand the effect of interaction terms on the chances to graduate, such as being an Asian female from a low-income status or a Black male from high-income status.
This research focusses on exploring hate crime data obtained from FBI's Uniform Crime Reporting Program which collects information on the occurence of hate crimes in states that voluntarilty submit their data.
For further details, read more or view the project!
... This is a personal project that I have been wanting to undertake for months. It will allow me to build on my data analytical, visualization, and reporting skills, while examining and analyzing events that bring U.S.A.'s systemic racism to light.
This research focuses on predicting the life expectancy of a nation based on various health features by applying regression models.
For further details, read more or view the project!
... We started by conducting basic exploration and transformations of the features to account for skewness. Then, we conducted backward elimination and applied a first-order model to see which variables are significant and found out that hepatitis B is not a significant variable. To make a more reliable and a better predictive model, we conducted step-wise regression on centered interaction efects between all variables. We found out that many of the interaction efects of the variables are significant proving that certain health factors have an infuence on life expectancy of a nation dependent on another health factor. However, our data set has a few clusters for various variables that also have large residuals and/or leverage points (points that have a large infuence suggesting that we should consider diferent transformations for these variables. Hence, while we have used the AIC criterion to make our predictive model more reliable, we would suggest conducting a few more tests and transformations to account for high leverage and residual points.
This data analytics project focuses on predicting income level (below or above $50,000/year) by applying machine learning models.
For further details, read more or view the project!
... The project was completed in Python using Jupyter Notebook. We began with exploratory data analysis looking at the variables, their distributions, and any anomalies like missing or duplicate values. We looked at correlation between the independent and dependent variables to understand possible impactful factors. The next step was to scrub the data and deal with all the anomalies, one-hot encode categorical variables, and scale numerical features. The third and last step of the process was to analyze and mdoel the data. We applied Decision Tree and KNN classification on the training and test sets to measure the predictive abilities of both these models. We found that the KNN classifier is the better model based on the applied metrics.
View ProjectThis analytics project focused on analyzing banknote images to see if the banknotes are genuine with staitistical analysis and fitting a logistic model (machine learning model) using R.
For further details, read more or view the project!
... We started by conducting basic exploration of each of the variables to understand if there were any skewness that needed to be dealt with. There was low to moderate skewness but it was not of a large concern due to which we did not conduct any transformations on our variables. Then, we conducted a first-order model to see which variables are signficant and found that all variables except entropy are signficant in determining if the image is of a real or fake banknote. After this, we conducted step-wise regression and got similar results because of which we added interaction effects and did stepwise regression with the AIC criterion and found that the interaction between variance and entropy, and kurtosis and entropy are significant at a 0.05 level. It appears that our model is a close-to-perfect fit of the dataset because it has linearity, as checked from model diagnostics, and high sensitivity and specificity rates indicating that its predictive powers is high. Further, the area under the ROC curve is close to 1 (0.9998), further confirmation of the model’s perfect fitness. There are points that have high leverage values and there is one point that has a high Cook’s distance value. For further investigation, we would suggest looking at these points to understand how we can deal with them.
View ProjectThis analytics project focused on predicting pollution levels (PM 2.5) based on air quality characteristics in Beijing by applying machine learning models using Python.
For further details, read more or view the project!
... The project was completed in Python using Jupyter Notebook. We began with exploratory data analysis looking at the variables, their distributions (using waffle charts), and any anomalies like missing or duplicate values. We looked at correlation between the independent and dependent variables to understand possible impactful factors. The next step was to scrub the data and deal with all the anomalies, one-hot encode categorical variables, and scale numerical features. We created two sets of training and test sets: with and without outlers and created dashboards to visualize the effects of removing the outliers on the model. The third and last step of the process was to analyze and mdoel the data. We applied multiple linear and polynomial regression to both sets of training and test data to find that the polynomial regression with outlers have better predictive ability.
View Project
Further details are provided on my LinkedIn profile
Further details are provided on my LinkedIn profile
Further details are provided on my LinkedIn profile
Further details are provided on my LinkedIn profile