<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Machine Learning on James's Peredutions</title><link>https://www.jamesgibbins.com/tags/machine-learning/</link><description>Recent content in Machine Learning on James's Peredutions</description><generator>Hugo -- gohugo.io</generator><language>en</language><lastBuildDate>Sat, 16 May 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://www.jamesgibbins.com/tags/machine-learning/index.xml" rel="self" type="application/rss+xml"/><item><title>Fraud Detection Console using IEEE Dataset</title><link>https://www.jamesgibbins.com/ieee-fraud-detection-console/</link><pubDate>Sat, 16 May 2026 00:00:00 +0000</pubDate><guid>https://www.jamesgibbins.com/ieee-fraud-detection-console/</guid><description>&lt;h2 id="introduction"&gt;Introduction&lt;/h2&gt;
&lt;p&gt;The dashboard is hosted on Render&amp;rsquo;s free tier: &lt;a href="https://ieee-fraud-detection.onrender.com/" target="_blank" rel="noopener"&gt;https://ieee-fraud-detection.onrender.com/&lt;/a&gt; (it can take a minute or so to load first time).&lt;/p&gt;
&lt;h2 id="the-data"&gt;The data&lt;/h2&gt;
&lt;p&gt;I used the IEEE-CIS Fraud Detection dataset from Kaggle: &lt;a href="https://www.kaggle.com/competitions/ieee-fraud-detection/" target="_blank" rel="noopener"&gt;https://www.kaggle.com/competitions/ieee-fraud-detection/&lt;/a&gt; Of all the fraud-related datasets I found, this seemed one of the most complete, with over half a million rows of data and hundreds of features.&lt;/p&gt;
&lt;h2 id="the-analysis"&gt;The analysis&lt;/h2&gt;
&lt;p&gt;As usual, I used a Marimo notebook. I imported the data and merged the transaction and identity data on the Transaction ID. I also added a column &lt;code&gt;has_identity&lt;/code&gt;, as only about a quarter of the transactions had identity data. This keeps identity coverage as an explicit signal, rather than hiding it inside lots of missing identity fields.&lt;/p&gt;</description></item></channel></rss>