Inspiration
Even a trip to our closest neighbor Mars is 480 million kilometers and takes approximately seven months. Instead of taking this long trip, why not stay on Earth and save a different galaxy: the Lego world. The problem is, some Lego worlds are more expensive than others. Thus, we wrote a program to find the most cost-efficient galaxy to save.
What it does
- Predicts the price of a Lego set based on piece count, age, difficulty, rating, and theme using a decision tree regressor. Now you can find the set which provides the most fun/dollar!
- Used program to summarize what characteristics made certain sets more expensive
Data from: https://www.kaggle.com/mterzolo/lego-sets
How we built it
We used pandas, NumPy to clean the data set before processing. We applied linear regression, decision tree regressor (highest accuracy), lasso using sklearn, completed cross-validation, and compared their scores. Used matplotlib to visualzie data for summary.
Challenges we ran into
- Dealing with null values: some columns contained 20% NaN. We replaced all NaN with the same value after analyzing the patterns in their price distribution.
- Running basic commands - first time using NumPy/pandas/matplotlib/sklearn = lots of compile errors
Accomplishments that we're proud of
- We can predict prices for lego sets!
What we learned
- Some basic commands in NumPy, pandas to clean datasets
- How to plot diagrams with matplotlib
- How to apply a model in sklearn
Log in or sign up for Devpost to join the conversation.