Data science specificially is about knowledge sharing and communicating what the data means to other people. I've personally spent countless hours pouring over hunt tables to decide where/when to hunt next year. Some analytical visual representations would certainly provide a more valuable solution. These will also lead to more questions about historical results.
This project has many variables that determine differing outcomes.
There are influences from multiple environments (supply/demand/regulations/weather).
This project also has an intuitive progression of the Analytics Roadmap.
This project is about working with interesting data. A lot of people do things with financial information or Twitter data; those can work, but the data isn’t inherently that interesting.
I started this back in 2015 which culminated into a Web-app example
This year I'd like to walk through the analytics roadmap and detail the nuances of each phase. The increase in value and difficulty as we progress through the phases should be pretty apparent.
I will demonstrate model building techniques common to Data Scientists by utilizing R. Its also important to note that data modelers have a sense of what they are modeling. Intuition and content expertise are incredibly valuable. In this case I'll note that I have been a Colorado elk hunter for decades, and am also familiar with the data provided from CPW after attempting to make sense of their hunt tables over the years.
CPW provides a lot of info for hunters to sift through. Let's start by accessing what they provide.
Additionally, weather can play a large role in hunting success. Let's grab some historical data from Dark Sky using their API
Goal I would like to hunt next year and would like to know which season will provide me the best chance of success for a certain Unit. How did things go in past years?
Any relationship between Units and their results? Are there other factors to consider?
- Preference points
- Draw results
- Herd size
- Weather
- Hunt season dates Goal What makes the first season in Unit 77 have the most hunter success? Why does it take less effort (hunting days) to be successful in the first season of Unit 77?
Use machine learning and predictive modeling to forcast future hunt seasons.
- I will perform preprocessing and data transformations
- Model building / tuning / training
- Model testing
- Release web service to analyze new data
- Predict the number of hunters in each unit,
then their breakdowns per season.
- Inputs could include Draw Results, historic transportation and lodging costs, median incomes, unemployment rates
- Predict the harvest in each unit, then their breakdowns per season.
- Input predicted number of hunters
- Additional inputs could include forecasted weather
- Predict the post hunt elk population
- Input predicted harvest results
- Additional inputs could include forecasted weather
- Use predicted values to generate success stats, etc
What can we influence? This is probably something CPW performs for future year’s regulations.