My last project of automating a data import pipeline for Chicago's crime data created the perfect environment for using past crime rates to predict future. I use SARIMA time-series forecasting to predict weekly crime rates 6-months out for the city and create a heatmap of location by time to further identify crime trends for Chicago.
more ...I pull data from the UCI Machine Learning Repo and use it to train a model which can identify advertisements based upon their image size and URL terminology. I work through cleaning the data, attempting a few different fitting algorithms, and end with some parameter-tuning of an ANN. My final model results in over 97% accuracy in classifying advertisements in testing. Originally conducted for a Machine Learning course as SHHU focused on the R language.
more ...As in the last post, here's some more work in excel with economic variables. This time I use value forecasts of 30y mortgage, unemployment, and personal income rates, figured in a similar manner as before (annual growth/change rates - 10y moving averages) to predict future levels of personal consumption expenditures. I run a multilinear regression analysis to forecast PCE based upon the three independent variables and end up with some pretty strong results and an adjusted R-squared of .974.
more ...I've wanted to get more practice with natural language processing, so I grabbed a dataset of Wikipedia comments from a past Kaggle challenge to attempt to classify toxicity of each comment. Train data was a collection of over 150k comments connected to user-defined classifications of 'toxic', 'severe toxic', 'obscene', 'insult', 'threat', 'hate'. Here's the process I took to create a model which identifies these particular classifications of future comments!
First is to import libraries we'll be using
Summary: It's been alittle over three years since I attempted my first ML project of classifying survival of the titanic.. A pretty famous dataset and task for this field I'd say. I achieved a accuracy of around 75% in my first attempt at this (I believe with a random forest model). I though it would be nice to revisit this and give it another go now that I'm a little more experienced. Additionally, I've wanted to explore ensemble learned some more, particularly XGBoost. Let's dive in!