Gbenga Ajiboye

A chain of coffee shops considered launching a new menu in certain locations of its chain. It ran the changes along few selected stores, over a 12 week period.

-> A weekly sales per store dataset was derived from general sales data.
-> Given the selected Treatment stores, two matches were derived per store using control variables.
-> A trend and seasonality estimate was generated to model what a typical demand trend within a year looks like.
-> The change in demand… Show more

A chain of coffee shops considered launching a new menu in certain locations of its chain. It ran the changes along few selected stores, over a 12 week period.

-> A weekly sales per store dataset was derived from general sales data.
-> Given the selected Treatment stores, two matches were derived per store using control variables.
-> A trend and seasonality estimate was generated to model what a typical demand trend within a year looks like.
-> The change in demand for the 12 week period was compared to the previous year's trend and seasonality
-> The estimated change in revenue per store per week was extracted, as well as other indicative variables of significance. Show less

A video game marketing arm is structuring its inventory management arm. The goal of this project is estimate over the next four months the expected demand, based on historical data.

-> The dataset was split into analytical and validation data sets, suing the more recent values for validation.
-> The monthly data was then passed into an ETS tool
-> The ETS terms were selected, and a forecast was run.
-> The forecast results were matched with the validation data set… Show more

A video game marketing arm is structuring its inventory management arm. The goal of this project is estimate over the next four months the expected demand, based on historical data.

-> The dataset was split into analytical and validation data sets, suing the more recent values for validation.
-> The monthly data was then passed into an ETS tool
-> The ETS terms were selected, and a forecast was run.
-> The forecast results were matched with the validation data set, in other to estimate the error.
-> An ARIMA model was also used to parse the data set.
-> The results from the ARIMA model was used to match the validation data sets.
-> The model with the least error was chosen, (i.e the ARIMA)
-> The model was used to forecast the next four month of expected sales Show less

A movie producer wanted to better understand film industry trends before releasing its next movie.

-> Data was extracted from a public movie dataset available on kaggle.com
-> The appropriate data preparation processes were performed using Alteryx
-> The transformed dataset was loaded into Tableau
-> A tableau Story and Dashboard was created, highliting genre success rates, using a Budget-Returns chart, to illustrate profitable genres and titles.
-> Franchises… Show more

A movie producer wanted to better understand film industry trends before releasing its next movie.

-> Data was extracted from a public movie dataset available on kaggle.com
-> The appropriate data preparation processes were performed using Alteryx
-> The transformed dataset was loaded into Tableau
-> A tableau Story and Dashboard was created, highliting genre success rates, using a Budget-Returns chart, to illustrate profitable genres and titles.
-> Franchises were shown to have a high Profitability position Show less

A bank trying to cope with a recent increase in loan applications, was in need of a BI Analyst to select applicants with a possibility of default within 0.99 to 0.50

-> The bank provided the dataset of the applicants, as well as previously repaid and unpaid loans from their own database
-> The dataset was cleaned out with irrelevant fields removed
-> Variables that were replicant variables were also removed
-> A holdout sample of 30% was set aside for… Show more

A bank trying to cope with a recent increase in loan applications, was in need of a BI Analyst to select applicants with a possibility of default within 0.99 to 0.50

-> The bank provided the dataset of the applicants, as well as previously repaid and unpaid loans from their own database
-> The dataset was cleaned out with irrelevant fields removed
-> Variables that were replicant variables were also removed
-> A holdout sample of 30% was set aside for validation
-> The resulting dataset was passed through a decision tree, Boosted Model, Logistic Regression Model, and Forest Model
-> Each model was tested with the validation set to assess the strength of each predictive method.
-> The model with the highest success raate, (the Forest Model), was chosen.
-> The new applicant dataset was then used piped through the model
-> Results were sorted from highest to lowest probability
-> Results less than 0.5 were dropped from the dataset Show less

A pet store chain is selecting the location for its next store.
Work Done:
-> Built a data set from disparate Data sources
-> Formatted and Cleaned extracted data
-> Performed a correlation analysis on determinant variables, to remove variables that are most similar
-> Built a profit model using determinant variables
-> Drew up a summary of expected Profit from each location
-> Selected Locations in top 50 percentile.

Creating an Analytical Dataset from a SQL Database

-> Using complex JOIN statements to retrieve data from a Supply Chain Database
-> Implementing subqueries to return comparative results from the Supply Chain Database
-> Presenting results using a Presentation in Powerpoint

A jewelry company wants to put in a bid to purchase a large set of diamonds, but is unsure how much it
should bid.
-> A regressive model was built using variable attributes to estimate the potential bid for each diamond, which was then summed to get a composite value for the bid of the diamonds