When MoMA was founded in 1929, its collection only featured a handful of nationalities. In 2016, MoMA acquired works from 53 different countries. This Tableau heatmap visualizes how MoMA acquisitions diversified over time.
How much money can we save if we are able to Predict weekly gas price trends? This project compared three models: 1) multivariate rolling regression, 2) ARIMA, and 3) logistic regression.
Different cuisines use different staple ingredients. This classification model labels recipes according to ingredients. The underlying techniques are bag of ingredients, logistic regression, and random forest.
Images also have ‘frequency’ signals, just like music. The discrete Laplacian is a generalization of Fourier transform on images. This project designs a discrete Laplacian filter to smooth noisy curves while preserving length and texture.
Can we guess shapes and automatically fill holes in drawings? This program constructs a probability distribution of gradients, and fill holes with simulated annealing.