Choosing the Best Ski Resort

This project functions to help mountain jockies (such as myself) choose the North American ski resort that's right for them. Whether they're looking to save the most money, find the mountain with the highest number of trails to explore, or find the mountain with the highest elevation (thus, the best snow), this program can help you find that hidden gem.

Instructions for Running this Program

1. Install Python 3
   • This program's interpreter is Python 3+, I am using Python 3.10.12, you can download and install Python 3 from this link.
     • Note: That link takes you to the Python 3 release page for Linux/UNIX Operating Systems. I highly recommend this program is ran on such system.
2. Install PIP: PIP is required for this program.
   • Windows Installation Guide:
     1. On your system search, search for Windows Powershell, Git Bash, or CMD. Right-click the result and click 'Run as Administrator'
     2. type and enter this command: curl https://bootstrap.pypa.io/get-pip.py -o get-pip.py PIP should download to your system.
     3. To install, enter this command: python get-pip.py
     4. Verify PIP installed by entering: python -m pip help Entering this command should display the location on your system of where PIP is installed
     5. Add a PATH variable for PIP
        1. Open the Windows Search, and type and enter "Environment Variables"
        2. System Properties should open, at the bottom of the window, click "Environment Variables".
        3. This will open a list of environment variables, double-click 'Path', or single-click and then click 'Edit'
        4. Click 'New', and then add the directory of where PIP is installed on your system. This directory should've been displayed from entering into your commmand prompt: python -m pip help
        5. Click 'OK' and the changes should save.
     6. Open a clean CMD, Bash, or Powershell, then type and enter pip help. This should display the same location information from step 4. You might have to instead enter pip3 help. If you're having issues, it might be wise to add the directory where your Python3 installation is located to the same PATH variables from step 5.
   • Linux/UNIX Installation Guide:
     1. In your system terminal, type wget https://bootstrap.pypa.io/get-pip.py This will download the installer.
     2. Install PIP to your system by typing and entering: python3 ./get-pip.py PIP should now be installed.

Unix/Linux Users Only

3. Configure the Shell Script.
   • The shell script located in SummitSelect/src will need a couple commands entered into the terminal before it will be an executable.
     1. CD into /SummitSelect (cd path/to/file/SummitSelect)
     2. Type and enter chmod +x run_code.sh and enter your password if/when prompted.
     3. run_code.sh should now be executable. Type and enter ./run_code.sh to run the shell script.
     4. Additionally: You yourself might have to edit the shell script, depending on where in your system your python interpreter is located. If you must, you'll be editing line 3. Important: Make sure you're in ./src if you're going to run the .py files manually.

Resources Used

Original Dataset

https://www.kaggle.com/datasets/ulrikthygepedersen/ski-resorts/code

North American Ski Resort Price Data:

https://www.onthesnow.com/north-america/lift-tickets

Pandas

https://pandas.pydata.org/docs/user_guide/10min.html

Data Pre-Processing

https://www.geeksforgeeks.org/data-preprocessing-machine-learning-python/#

Developing a Ranking System

https://www.stratascratch.com/blog/methods-for-ranking-in-pandas/

• Referenced Code From: https://github.com/Didayolo/ranky/blob/master/ranky/ranking.py

Weighted Sum Model

https://www.geeksforgeeks.org/weighted-sum-method-multi-criteria-decision-making/ https://www.youtube.com/watch?v=iAr4tGEYQZ0&ab_channel=HigherMathematicsLearning https://www.youtube.com/watch?v=EGf9bUgl3R8&ab_channel=labmao

Merging Dataframes

https://pandas.pydata.org/docs/user_guide/merging.html

Argument Parsing in Python Using argparse

https://www.geeksforgeeks.org/command-line-option-and-argument-parsing-using-argparse-in-python/ https://www.youtube.com/watch?v=88pl8TuuKz0&ab_channel=NeuralNine

Additional Aid

• I utilized generative AI to assist with debugging and researching topics.

Project Report

To re-iterate, this project serves to curate a list of top ski resorts tailored to the preferences of the user.

Data Mining Task

The task of this project is to answer a series of questions about various ski resorts across North America, and curate a list, ranking these ski resorts tailored to the answered questions. The questions are as follows:

1. Which ski resorts contain the most runs?
2. Which ski resorts are the most affordable?
3. Which ski resorts have the highest elevation (and consequentially, the best snow)?

Using the two datasets listed under 'Resources Used', and some tedious browsing on Google, I've compiled three separate datasets, each containing data for the same list of 100 ski resorts across the United States and Canada; run count, full-day adult lift ticket prices, and the peak elevation of the resort, respectively. The overall goal is to use the three datasets and the three listed questions to compile a list of ski resort recommendations, of which are the "best bang for your buck", or not, if you decide you want to find and attend the most expensive ski resort. Someone using this program could give the prompt that they want to find the ski resorts with the worst pricing, the least amount of runs, and the lowest elevation if they so please!

Methodology

In order to get the data ready for the main algorithm used in this program, the weighted sum model, I needed to perform rather intensive data preprocessing, followed by a ranking system for each key feature. To keep it brief, the data preprocessing, which is mostly done in process_data.py, normalizes the data before merging them into a new dataset. This new dataset is then used as input for the ranking functions in ranking_data.py. The ranking system of this project is designed to take in user input for the three questions listed under 'Data Mining Task', and rank the three unique criteria based on the users' responses. To be specific, the user is asked to give a basic 'yes' or 'no' response to each question. For example, a 'yes' to run count would rank Run Count from most to least runs per resort, a 'yes' to prices would rank Price (USD) from least expensive to most expensive lift tickets, and a 'yes' to elevation would rank Peak Elevation (m) from highest to lowest.

Once the new input file is created, we come to the real meat and potatoes of this program, the Weighted Sum Model, or also known as the Multi Criteria Decision Making Model. This model takes in multiple possible pieces of data that relate to a list of items, as we have in our dataset (run count, price, elevation) for each resort, and determines the best item based on the order of scoring for each criteria. The score for each criteria is used to compute a total overall score, and that is how, in this program, each resort is ranked in the final list that is written to the output file.

Weighted Sum Model

$A_i^{WSM-score} = \sum_{j=1}^n w_j a_{ij}$, for $i = 1, 2, 3, ..., m$ Where m represents the different features, and n is the decision criteria. $w_j$ denotes the weights and $a_ij$ denotes the performance value, which we substitute with the results of z-score normalization.

Results

First 10 Resorts in Post-Data Pre-Processing

• We can see in these results, the values of each piece of criteria have been normalized and prepared for ranking.

First 5 Resorts After Each Criteria Has Been Ranked (Based on a 'Yes', 'Yes', 'No' Set of Preferences, Respectively)

First 5 Resorts From the Final Output

• This has been taken directly from Ski_Resorts_Results.txt, while the individual rankings were taken directly from console output.
• In our final result of a 'yes', 'yes', 'no' set of user preferences, the largest factor in ranking the resorts is found to be pricing.
  • So while we do want to find resorts with a higher number of runs, wanting to achieve the lowest possible price takes precedence.
• Something about this which I would like to note, this result might differ if I took a different approach with taking in user input.
  • I think if I were to instead have user input be on a spectrum (Say for Price, the user can enter 'Absolutely', 'Yes', 'Average', 'Don't Care', 'Don't Care At All), and have the same type of answering for the other criteria, then perhaps the results would be less bias towards another.

Issues With the Results

• I do not believe the results are entirely accurate. When reversing the data from their normalized scores, Sandia Peak being ranked #1 doesn't seem entirely accurate.
  • With the prompt I gave, I would expect a resort that has a high number of runs, at a lower price, with a lower elevation, but Sandia Peak has one of the lowest run counts on the list, and a relatively high peak elevation, so the only accurate result is its very low price.
  • My thinking is that the normalization has unintentionally created an extreme bias when computing the weighted sum, as resorts with relatively different values in run count, have been normalized to perhaps be too close to being equal, thus cannot be ranked properly. However, I am still chopping away at this minor issue, so I cannot say accurately as of now if this is the problem. Overall, the technical goals of this project have been achieved.