A collection of exploratory data analyses
2016 Presidential Votes & Taxes
An analysis of voting results in the 2016 presidential election vs. state populations, tax payments, and federal subsidies received.
https://github.com/dbolta/exploratory_exercises/blob/master/2016_presidential_votes_vs_taxes.ipynb
Bayesian Inference Visualized
Bayesian inference simulations and visualizations inspired by Rasmus Baath's datacamp course.
https://github.com/dbolta/exploratory_exercises/blob/master/Bayesian%20Inference%20Visualized.Rmd
https://github.com/dbolta/exploratory_exercises/blob/master/Bayesian%20Inference%20Visualized.ipynb
Hot Ones Spice Graphs
A trellised visualization to determine how much of a shock each proceeding sauce is for the contestant.
https://github.com/dbolta/exploratory_exercises/blob/master/hot_ones_graphs_R_kernel.ipynb
Hours of Work to Pay for a Hospital Visit
Utilizing Data.World's Monday Makeover mortage and earnings data with Kaiser Family Foundation data on Medical Costs
https://github.com/dbolta/exploratory_exercises/blob/master/Hours%20to%20Pay%20Hospital%20Visit.ipynb
Prescriptive Fit Data with Relu
Perfectly fit 1D data with single hidden layer using relu activation.
https://github.com/dbolta/exploratory_exercises/blob/master/Prescriptive%20fit%20for%201D%20data%20using%20relu.Rmd
What does it mean to be masculine?
Analysis if Nate Silver 538.com masculinity survey data
https://github.com/dbolta/exploratory_exercises/blob/master/538_man_survey.pdf
https://github.com/dbolta/exploratory_exercises/blob/master/538_man_survey.Rmd
Baby Tylenol Dosing
Calculate exact dose of acetaminophen based on baby's weight
https://github.com/dbolta/exploratory_exercises/blob/master/baby%20tylenol%20dosing.ipynb
Baby Weight Bayes
Bayes and MCMC probability distribution of baby's weight via WHO growth chart (prior) and measurements at doctor (likelihood)
https://github.com/dbolta/exploratory_exercises/blob/master/baby%20weight%20bayes%201.ipynb
Cauchy Distribution
Comparing different ways to calculate the Cauchy Distribution, visualized with Altair
https://github.com/dbolta/exploratory_exercises/blob/master/Cauchy%20Distribution.ipynb
Continued Fractions
Computing irrational numbers (the infinity one level above counting numbers)
https://github.com/dbolta/exploratory_exercises/blob/master/Continued%20Fractions.ipynb
Euclid Greatest Common Denominator
Implementation of the euclid method to find gcd
https://github.com/dbolta/exploratory_exercises/blob/master/Euclid%20GCD.ipynb
Galveston Bay water readings around Hurricane Harvey
Review of water sampling data and how it correlates with Hurricane Harvey.
https://github.com/dbolta/exploratory_exercises/blob/master/Galveston%20bay.ipynb
Houston Floods
Use count of Houston flood days over 20 recent years to do Poisson and Bayesian timing analysis.
https://github.com/dbolta/exploratory_exercises/blob/master/houston%20floods.ipynb
Infinite Resistors
Quick little infinite sum calculation. Inspired by xkcd
https://github.com/dbolta/exploratory_exercises/blob/master/Infinite%20Resistors.ipynb
LSTM vs GRU in keras
Predict sine wave using window function, then compare which RNN is more accurate.
https://github.com/dbolta/exploratory_exercises/blob/master/predict_sine_LSTM_vs_GRU.ipynb
Monte Carlo Integration
Use Monte Carlo and MCMC to Integrate
https://github.com/dbolta/exploratory_exercises/blob/master/Monte%20Carlo%20Integration.ipynb
Neural nets from scratch with numpy
4 neural nets from scratch:
1. Multilayer with relu activation
2. #1 with additional hidden layer
3. #2 with bias
4. #3 with sigmoid activation
https://github.com/dbolta/exploratory_exercises/blob/master/numpy_nnet_1.ipynb
Neural nets with pytorch
Deep learning with weight perturbation between epochs
https://github.com/dbolta/exploratory_exercises/blob/master/pytorch_practice_1.ipynb
Probability Examples
Originally inspired by Probability, chapter 5 of Intro to Algorithms
1. Birthday paradox
2. Coupon collector's problem
3. Poisson and exponential events probabilities
4. Goodness of fit: chi-squared and k-s tests
5. Jeep problem
6. Streaks: hitting slumps, NBA shooting hot streaks
7. Streaks with impurity: hitting 4-out-of-5, shooting 15-out-of-20
8. Simulated Chinese Restaurant Process
9. Simulated Dirichlet Process
https://github.com/dbolta/exploratory_exercises/blob/master/probability%20examples.ipynb
Power of a Study with Bokeh
Bokeh version of ipywidgets functionaliy
https://github.com/dbolta/exploratory_exercises/blob/master/Is_My_Study_Under-Powered_5e_linked_tabs.py
Power of a Study with ipywidgets
Power analysis of an improvement study.
1. Power given sample size.
2. Needed sample size to reach desired power.
https://github.com/dbolta/exploratory_exercises/blob/master/Power%20of%20Study%20ipywidget.ipynb
Tensorflow 2.0
Custom training to perfectly fit a 2D input set.
https://github.com/dbolta/exploratory_exercises/blob/master/Tf_2.0_sequential_training.ipynb
Trump Fact Checking
Poisson analysis based on Washington Post's Fact Check summary statistics.
https://github.com/dbolta/exploratory_exercises/blob/master/Trump%20Fact%20Checking.ipynb