What are the most under-rated hiking destinations?
Using data collected from the web, apply exploratory data analysis to find hiking destinations that are under-rated.
Note: From the outset, I had to define what "under-rated" means. I concluded that I wanted places that have one or multiple of the following:
- A relatively high amount of hiking trails
- A relatively high amount of hiking trails labeled "Moderate" or "Difficult". This is because many parks and city walking areas are classified as trails- with a label of "Easy". If I filter out "Easy", then I would likely be left with a more rural-centric selection.
- A small overall tourist industry, or a small percentage of tourism.
Visit the link to the github page here: https://github.com/dgoldman916/low-key-great-hikes
Step 1: Collecting the Data
- To find destinations that are of interest to travelers, I collected Tripadvisor's list of popular US destinations. This list includes hundreds of locations, some are places with an emphasis on outdoors experiences, others are towns or big cities.
- Afterwards, I brought the data into https://www.alltrails.com, a repository of hiking trails by location. With my list from Tripadvisor, I gathered information about the trails for each location (This includes each trail name, its rating, number of reviews, difficulty, and the total trails returned for each location.
- Looking for under-rated destinations, I conjectured that those places would have a moderate to small tourist industry, or that their tourist industry in relationship to their trail count and popularity would be sortable, in a way that can separate hiking destinations from non-hiking destinations. In order to add this dimension to the data, I sent my list of locations to https://www.whatsthecounty.com, and saved the results of each county. I also collected US Census Data about economy sizes by county.
Step 2: Preprocessing the Data
- To make a database that includes a full set of data for each location I joined my collected files into a pandas data frame. You can check out https://www.github.com/dgoldman916.com/underrated-hiking for the code.
Step 3: Exploratory Data Analysis
- With data about tourist industry size, total industry, hiking trails, reviews, etc. I sorted the data on multiple columns. The foremost rows were locations with prominent outdoor features: Aspen, Colorado or Moab, Utah. Further down the list had to live my under-rated locations. But how do they get to the top of the list?
- I decided to use the available data to score each location on its features, and use those scores to sort the data, as well as visualize them. With some basic Exploratory Data Analysis like groupby, aggregating sums and means, it was possible to quantify a hiking locations average rating score, total reviews, whether or not the location had mostly hard trails (usually something like mountains) or easy trails (usually city parks). See the github links in Step 2 for examples of the functions built to perform EDA on the data.
Step 4: Expanding
- At this point, I was building functions that added columns to the dataset, and visualizing them inside of a jupyter notebook. To make this part of exploring data easier and more robust, I built an app that automatically plots and sorts the datable. It creates an easier feedback for the process, and is a useful tool for sharing. Visit below:
Step 5 Refining
- The previous steps succeeded in gathering hiking-friendly locations from Tripadvisors list. In the visualization, it is easy enough to explore and sort locations based on any of their features. However, there is no definitive scale for under-ratedness, although there are included some systems of scoring. This presents interesting challenge questions:
- What additional features contribute to under-ratedness?
- Can a user-defined or user-directed definition of under-ratedness be incorporated into the app?
- Can a user-inputted location return an under-ratedness score?
- I plan to address these questions in future iterations of this project. Some of my plans include:
- Collect data from a larger repository than Tripadvisor
- Collect review texts, and key words associated with each location's hiking trail results. Potentially perform a sentiment analysis.
- Source more features per location, some could include:
- Population, population density, urbanization, climate
- Transportation, gas stations, roads, incorporate elements into an "ease-of-access" index
- Flora and fauna count, other attractions.
- Get toll payments from national parks, other indicators of specific trail usage.