Discussing proper (and improper) ways to perform k-fold cross-validation on datasets — The k-fold cross-validation is a popular statistical method in machine learning applications. It mitigates overfitting and enables models to generalize better with training data. However, in practice, the technique can be trickier to execute compared to the typical train-test split. If used incorrectly, the k-fold cross-validation can cause data leakage.

An end-to-end data analysis project for examining how the pandemic has affected US air travel — Introduction The Covid-19 pandemic has had an unprecedented impact on many economic sectors. Among these sectors, the aviation industry especially deserves attention due to its considerable influence on the prosperity of individuals, businesses, and countries alike. Thus, I believe it is worth digging deeper into this subject to answer questions like:

How to ensure that missing data is dealt with correctly — Missing values are a dilemma in many datasets, but many users are eager to allocate less time to deal with them so that they can focus more on modeling. I write this after recently running into many portfolio projects that treat missing data as an afterthought. There is a lot…

Uncover what this statement does and why you should use it in your own scripts — Starting out with Python, you’ve likely written code in the following format: def function(): pass function() In other words, your operations typically entail creating a function or class and then calling it. Simple, right? If this applies to you, you might be a bit confused when you see other scripts…

An end-to-end data analysis project for examining how the pandemic has affected US air travel — The Covid-19 pandemic has hampered operations in virtually every industrial sector in the US. The influence of the pandemic is especially evident from the downturn in activity in the aviation industry. …

How you can (and why you should) create custom transformers — Scikit Learn transformers (not to be confused with deep learning transformers) are classes in the Scikit Learn package that facilitate transformations in given datasets. These transformers can carry out various operations like normalization and principal component analysis. …

Overview and implementation with Python — Many machine learning algorithms perform better when they are trained with discrete variables. For that reason, it is common for feature engineering procedures to incorporate discretization. You may find that many of the winning submissions in machine learning competitions leverage this technique. Don’t be too intimidated by the 5-syllable term…

Why high-dimensional datasets are so undesirable — The curse of dimensionality plays a substantial role in how we transform our data prior to any modeling. That being said, with all the negative sentiment toward high-dimensional datasets, it’s worth asking how incorporating too many predictive features can sabotage model performance in the first place. After all, features provide…

A look into a tool that allows a pain-free troubleshooting experience — Do you get tired of constantly dealing with pesky bugs? Even with the debugging features provided by the popular IDEs, the smallest of errors can take precious time to troubleshoot. What if I told you that you can have a better experience with dealing with errors by proactively performing sanity…