mhoshko
commented
4 years ago Data Preparation and Modeling (18 out of 20%)
The code was cleaned as much as possible, however the dataset itself still had some missing areas that could have potentially been addressed.
Validation and Operationalization (18 out of 20%)
I did use cross validation and discussed the operationalization of the project, however, I could be creating a prediction that is not going to work how I predict with the addition of emergency response time data and I never discuss the possibility of it not working.
R Proficiency (20 out of 20%)
I worked to document the process well, and keep the view for readers clean through appropriate chunk naming.
Communication (18 out of 20%)
My wording throughout can be more verbose than necessary and could be less repetitive but the result is communicated properly.
Critical Thinking (20 out of 20%)
I clearly state the action that should be taken to continue with this model to make improvements on creating a stronger relationship to expose problems with the opioid epidemic in Connecticut.
Summary
The project looked into the opioid crisis with a focus on the drug Fentanyl and a location of Connecticut. It looked into how Fentanyl played into the crisis over the past 7 years as well as the correlation between the location of the overdose, and the location of the death for the person who overdosed. The conclusion was that Fentanyl overdoses are on the rise and that the state of Connecticut needs to take some type of action in order to decrease the number of overdoses.
Data Preparation
The tables that were built contain information about drug overdoses in Connecticut. It tells information on where/when/how it happened, including the location, date, and time of death/overdose. It also tells you what drugs were found in their system at the time of death.
The portfolio is well organized and tidy. The data is cleaned so that it is easy to access and work with, along with an ease of understanding the columns. It is very easy to follow what changes are being made, and the changes are documented and shown each and every step of the way to ensure a better understanding of how the portfolio and data are progressing.
Modeling
The main focus of the models were to determine which variables play the biggest part in determining the location of the death compared to the location of where the overdose occurred. The variables that showed the most correlation to this were the city where the overdose occurred, and the type of location it occurred in. The model is interpreted correctly in that the confidence is not high enough to accurately predict the location.
Validation
The model has been cross validated and a graph was supplied to help visualize the differences between the models predicted output and the correct output supplied by the data.
R Proficiency
There were a large number of R functions used that I had never seen before, but proved to be very helpful in progressing the portfolio's goals. The code is very easy to understand due to how it was broken up into different segments and those segments each had their own descriptions about what the code was doing.
Communication
Each step of the way was described in a way that made understanding it simple, while still accurately providing explanations as to what was happening. The were a lot of visualizations supplied that gave you a good sense of what the data was and how all of the different variables related to each other.
Critical Thinking
The future of the project was described as needing a few more data points before it can be determined if there is a need to take action on the opioid crisis, specifically Fentanyl. Some things that could also play a role in this is the ease of access/amount of supply of the drug.