codyclonts / animal_crossing_villager_popularity

animal_crossing_villager_popularity

Animal Crossing New Horizons Character Popularity Analysis

Project Oversight

• Animal Crossing: New Horizons (acnh) is video game in which the player controls a character who moves to a deserted island. In this game, the player accomplishes assigned tasks, develops the island as they choose, and brings other characters to live on their island. For this project, I am using information from the Animal Crossing New Horizon character catalog csv to acquire, prepare, explore, and create models to predict whether or not a new character will be successful.

  • A successful character is defined as falling within tier 1, 2, or 3 in the popularity poll. This accounts for about 17% of the characters.

• This information is based off of a popularity poll that is updated monthly. It is current as of June 30, 2022.

Business Goals/ Description

• Find key drivers of success for characters in the game Animal Crossing New Horizons.
• Construct a machine learning model that can be used to predict whether or not a character is successful given the characteristics of the character.
  • The models that will be used are decision tree models, random forest models, and k nearest neighbor models.
  • The top performing model should be able to beat a baseline accuracy of 82%.
• This information will be used to give more insight to what characteristics should be given to characters in order to:
  • Develop characters that have a high likelihood of being successful in order to market those characters and sell them in bundles.
  • Reduce manpower hours spent developing characters that will not be successful.

Project Goals

Answer the questions: What features, if any, can be used to help determine whether or not a new character will be successful in the game Animal Crossing New Horizons? How can these features be combined to predict the success of a new character?

• Ask questions during the exploration phase to better understand what characteristics could be factoring into the target variable of whether or not a character is successful.
  • Answer those questions with statistical testing and visualizations.
• These answers will be used to help predict whether or not a new character will be successful in the future.
• Construct machine learning models to predict whether or not a new character will be successful.
  • Run the most effective machine learning model against test.
• Make final recommendations and provide next steps.

Executive Summary

• There are a few things that can help in determining whether or not a character will be successful in Animal Crossing New Horizons.

• Some of the things that we can use to help predict whether or not a character will be successful are (in order of importance) :

  • Species, Style_1, Gender, and Personality
  • A character's hobby can also be a slight driver in determining whether or not they will be successful, but it isn't as strong of a driver as the aforementioned categories.

• The decision tree model that I produced is able to determine whether or not a new character will be successful or not with 96% accuracy and with a 70% recall.

Audience

• Peers

Deliverables

• Jupyter Notebook containing the final report.
• Python Modules that can be used to reproduce the work.
• Scratch notebook that can be referred to for my work.
• Readme file explaining the project.

Data Dictionary

Project plan

• Acquire the animal crossing new horizons character csv file from https://www.kaggle.com/datasets/jessicali9530/animal-crossing-new-horizons-nookplaza-dataset
• Clean and prepare the data for the exploration file. Create a wrangle.py file to recreate the work.
• Explore the data and ask questions to clarify what is actually happening.
  • Ensure to properly annotate, comment, and use markdowns.
  • Write out each null and alternative hypothesis.
  • Visualize the data.
  • Run statistical test on the data.
• Create different machine learning models.
  • Decision Tree Model.
  • KNN model.
  • Random Forest model.
• Choose the model that performs the best.
  • Evaluate on test.
• Deliver final presentation to peers.

Initial Hypotheses:

• The species, gender, hobby, and color_1 are key drivers of whether or not a character will be successful.
• While favorite_song might be a driver, I don't know that it will be significant enough to determine whether a character will be successful.
• The style of the character will also be a driver of character success as it influences the characters visual appearance.
  • The style_1 and color_1 columns combined are generally used to determine the visual appearance of the characters in the dataset.

Reproduce the Project

• In order to reproduce the project, you will need:

  • A Copy of the csv from this repository.
  • Ability to clone this repository to local environment.
  • Ability to run the notebook.

• If you would like to recreate the work with up to date information, you will need to create an updated csv by doing the following:

  • copy the villagers.csv from https://www.kaggle.com/datasets/jessicali9530/animal-crossing-new-horizons-nookplaza-dataset
  • Access https://nookipedia.com/wiki/Villagers/New_Horizons in order to get the information for the missing characters
  • Access https://www.animalcrossingportal.com/games/new-horizons/guides/villager-popularity-list.php#/ in order to get the ranking position for each of the characters.