Hi, I’m Migs Germar.
Welcome to my data science blog. Feel free to explore my posts below. I recommend checking out My Portfolio Projects, which lists my best works. You can also click a category on the right side of the page to filter the posts.
If you’d like to get in touch, see the About page.
Thanks for visiting!
Using a Neural Network to Classify Handwritten Digits
I compare the performance of a neural network to that of a K Nearest Neighbors model in classifying images of handwritten numbers. I also demonstrate the use of Grid Search for hyperparameter optimization.
Comparison of Regression Models for Predicting Bike Rentals
I compare the performance of three machine learning models (linear regression, decision tree, random forest) in predicting the number of bike rentals that may occur at a given time in Washington, D.C.
Using Linear Regression to Predict House Sale Prices
I interpret linear regression results to determine features that significantly affect house sale prices. I then use the same model to predict house prices, and evaluate the model using stratified k-fold cross-validation.
agriHanda: an Agricultural Disaster Risk Web App
I developed a web app that won two awards in the Project SPARTA PH Open Data Challenge for Butuan City.
Answering Business Questions for an Online Music Store using SQL
I use intermediate SQL techniques like views, joins, aggregations, and set operations in order to solve 4 scenarios about a hypothetical online music store. Results are communicated with Matplotlib and Altair visualizations.
How Student Demographic Affects SAT Performance in NYC
I combine multiple datasets about NYC high schools to explore the effect of students’ demographics on SAT performance. The project ends with a multiple linear regression model showing that the effect of race percentages on a school’s average SAT score is significant.