Pokemon Analysis
Statistical Techniques
We chose to use multiple linear regression to predict our target variables. Although the barebones linear regression models were working well, we were able to improve the results of our models by utilizing various statistical techniques such as the Box Cox Transformation. We also used a package called SignifReg which helped us build models with the most relevant variables.
Interesting Results
Both of our final models were able to explain over 50% of variability in our target variables based on attributes. Although these results are good, there is a still a question of what factors outside of a Pokemon's attributes contribute to height and weight.
Predicting Weight & Height of Pokemon
For this project our group utilized the Pokedex, which is a catalog that contains all Pokemon and their relevant attributes and descriptors. Using these variables, we sought to build predictive models that could guess the weight and height of a Pokemon based on their attributes.