bryce-bowles

alchemy-broker-modeling

Performed segmentation analysis and predictive modeling on insurance broker performance to conclude a random forest model (highest AUC of 73%) predicted whether 2020 Gross Written Premium will increase or decrease from 2019 with a misclassification rate of 35%. Four classification models (classification trees, logistic regression, random forests, and support vector machines) were built, evaluated, and then tuned for prescriptive measures to analyze broker performance. Explored, visualized, and described five groups of brokers using principal component analysis.

diet-and-manufacturing-optimization

Diet Problem and Manufacturing Problem: Decided how much of each of each dessert to consume per day so that taste index is maximized, and calories and grams of fat are minimized, subject to constraints (Algebraic Formulation).

KJ-manufacturing-TSF

KJ Manufacturing Company case scenario: Discussed the forecasting process at KJ Manufacturing, any relevant factors about the company and industry that are pertinent to the new forecast and Ken’s forecast. Forecasted monthly revenues for KJ Manufacturing for the coming year. Used a variety of methods and graphically displayed them. Explained and supported the new forecasting approach as well as the choice of models and the rational for parameters selected. Prepared a report to owner explaining and supporting the forecast.

lending-club-classification

Built a logistic regression model and a classification tree model for predicting the final status of a loan based on various variables available. Confusion matrix and misclassification rate for each model for a test dataset. Variables that appear to be important for predicting outcome. Plotted and described the ROC curves and AUC for the four models.

lending-club-pca-cluster

Performed a Kmeans cluster analysis to identify 7 groups or clusters of the borrowers by income, loan amount, employment length, home ownership status, and debt-to-income ratio. Included Data Preprocessing and Removing Outliers.

office-workspace-optimization

Won Optimization model class competition issued by Dr. Brooks (M.D.A. department chair and professor). Proposed Python, Pyomo and GLPK network optimization model approach with binary variables and logical constraints to simulate reorganization of 1700 workspaces across 17 floors, while allocating for changing project teams and requirements. Provided report to IT Vice President, Christine Holzem at the Federal Reserve Bank of Richmond.

TS-Apple_Watch_Data

Time series health workout data was extracted from my Apple watch to analyze workout variables. A Scope, descriptive statistics, pivot tables, C-Chart and scatter plots were created to check workouts outside of control. Tableau work was used to display correlations.

ts-plots.R

ts-plots.R

TS_PBS_Australia_perscription_data

.r Time series plots using data on Australia Medicare prescriptions

tsf-richmond-bank

Using MS Excel and R, accurately forecasted total core deposit data from a Richmond Bank. The Holt’s Linear Exponential Smoothing had the overall lowest “Quick and Dirty” MAPE (1.2%), the lowest overall Maximum MAPE (3.49%), and consistently more accurate projections for each of the forecast horizons. Overall, the Unaided, Holts Linear Exponential Smoothing, and both regressions overestimated while the Naïve, 12 Month (M) Center Moving Average (CMA), 3M Moving Average (MA), 6M MA, Damped Trend Exponential Smoothing, and Simple Exponential Smoothing underestimated.

bryce-bowles/office-workspace-optimization

Won Optimization model class competition issued by Dr. Brooks (M.D.A. department chair and professor). Proposed Python, Pyomo and GLPK network optimization model approach with binary variables and logical constraints to simulate reorganization of 1700 workspaces across 17 floors, while allocating for changing project teams and requirements. Provided report to IT Vice President, Christine Holzem at the Federal Reserve Bank of Richmond.

bryce-bowles/alchemy-broker-modeling

Performed segmentation analysis and predictive modeling on insurance broker performance to conclude a random forest model (highest AUC of 73%) predicted whether 2020 Gross Written Premium will increase or decrease from 2019 with a misclassification rate of 35%. Four classification models (classification trees, logistic regression, random forests, and support vector machines) were built, evaluated, and then tuned for prescriptive measures to analyze broker performance. Explored, visualized, and described five groups of brokers using principal component analysis.

bryce-bowles/diet-and-manufacturing-optimization

Diet Problem and Manufacturing Problem: Decided how much of each of each dessert to consume per day so that taste index is maximized, and calories and grams of fat are minimized, subject to constraints (Algebraic Formulation).

bryce-bowles/KJ-manufacturing-TSF

KJ Manufacturing Company case scenario: Discussed the forecasting process at KJ Manufacturing, any relevant factors about the company and industry that are pertinent to the new forecast and Ken’s forecast. Forecasted monthly revenues for KJ Manufacturing for the coming year. Used a variety of methods and graphically displayed them. Explained and supported the new forecasting approach as well as the choice of models and the rational for parameters selected. Prepared a report to owner explaining and supporting the forecast.

bryce-bowles/lending-club-classification

Built a logistic regression model and a classification tree model for predicting the final status of a loan based on various variables available. Confusion matrix and misclassification rate for each model for a test dataset. Variables that appear to be important for predicting outcome. Plotted and described the ROC curves and AUC for the four models.

bryce-bowles/lending-club-pca-cluster

Performed a Kmeans cluster analysis to identify 7 groups or clusters of the borrowers by income, loan amount, employment length, home ownership status, and debt-to-income ratio. Included Data Preprocessing and Removing Outliers.

bryce-bowles/TS-Apple_Watch_Data

Time series health workout data was extracted from my Apple watch to analyze workout variables. A Scope, descriptive statistics, pivot tables, C-Chart and scatter plots were created to check workouts outside of control. Tableau work was used to display correlations.

bryce-bowles/ts-plots.R

ts-plots.R

bryce-bowles/arima-r

ARIMA Model created with r.

bryce-bowles/bbowles.github.io

bryce-bowles/bryce-bowles

README

bryce-bowles/bryce-bowles.github.io

bryce-bowles/business-data-analytics

Terms, classification models, test and training dataset splits, logistic regression models, classification tree models, ROC curves, AUC, confusion matrix, support vector machines, variance, bias, leakage, MAE and RMSE, R squared, LASSO approach (penalty on the coefficients) etc.

bryce-bowles/car-loan-negotiation_goal-seek

Used Excel Goal seek to negotiate a car purchased with variables such as Price, APR, Years, Payment/month.

bryce-bowles/decision-tree-midterm

Probabilities, Decision Trees and Influence Diagram scenarios

bryce-bowles/differencing

Differencing using r.

bryce-bowles/doordash-strategic-analysis

In depth analyses on each: Industry Analysis, Environmental Analysis, Strategic Review, and Growth Through Acquisition.

bryce-bowles/GIS-fingerprinting

bryce-bowles/helpdesk-optimization-proposal

Proposed optimization and simulation framework to benefit helpdesk request distribution and simulate future request volume.

bryce-bowles/MDA_Course-info

MDA Course Info

bryce-bowles/mobile-munchies

Mobile Munchies is deciding how much of each type of juice to prepare for the week. Given the ingredients and cost, a python model using Pyomo and GLPK determined the optimal amount of each type of lemonade to produce so the profits maximized subject to the constraints.

bryce-bowles/opioid-prescribing-rates

Semester long project working with Virginia Department of Social Services to assist in data centric reengineer useful data into VA’s major FAACT database. Tableau dashboard analysis and presentation created using data from 2016 to 2019 on Medicare Prescribing rates.

bryce-bowles/Red-Tomato-gardening-tools

Red Tomato Gardening Tools and Sporting Goods Company. Demand forecast optimization problem model using Python, Pyomo and GLPK in Python. Multifactor objectives and constraints solved using algebraic formulation to allocate and minimize cost. Excel Solver used to allocate how much of each product to produce so that profit is maximized

bryce-bowles/scc-work-dbms

Completed and proposed an Automated Systems Database to Manager via SQL Server and MS Power BI Version. Centralized Relational SQL Database to help produce the appropriate roles for a position, creating consistency throughout departments and job titles (with the exception of optional roles for additional access) and reduce the number of access roles that are kept when changing positions. The DBMS unifies and consolidates system access to improve data security as well as onboarding and offboarding efficiency.

bryce-bowles/SCC-Work-Experience

Serve as a Data and Systems Analyst Liaison between the Bureau of Insurance (BOI) and Information Technology Division (ITD), assisting Automated Systems’ Manager and Chief.

bryce-bowles/taylor-clothing-dbms

Business rules, user requirements, ER diagram, entity relationships etc. (Oracle APEX)

bryce-bowles/ts-decomposition.R

ts-decomposition.R

bryce-bowles/ts-exponential-smoothing

ts-exponential-smoothing

bryce-bowles/ts-regression.R

Time Series Forecasting with Regression

bryce-bowles/whiskey-prediction

Logistic regression model to predict the best and worst whiskeys using Confusion Matrix with training and validation samples. Correlation matrix, goodness of fit statistics, Hosmer-Lemeshow test, chi-squared, confusion matrix, Scatter plots, box plots etc.