---
title: "Discussion on \"99% Accuracy Is Not Always Good: What Fraud Detection Taught Me About ML\" | Hashnode"
description: "Discussion on \"99% Accuracy Is Not Always Good: What Fraud Detection Taught Me About ML\". My fraud detection model hit 99% accuracy and still failed. Here’s what imbalanced data, precision, recall, F1-score, confusion matrices, and SMOTE taught me the hard way."
url: "https://hashnode.com/posts/99-accuracy-is-not-always-good-what-fraud-detection-taught-me-about-ml/6a82f9adc962ba2c8a3e2de7"
---

[Feed](https://hashnode.com/feed) /

Discussion

# [99% Accuracy Is Not Always Good: What Fraud Detection Taught Me About ML](https://godfident-data.hashnode.dev/99-accuracy-is-not-always-good-what-fraud-detection-taught-me-about-ml?utm_source=hashnode&utm_medium=feed)

My fraud detection model hit 99% accuracy and still failed. Here’s what imbalanced data, precision, recall, F1-score, confusion matrices, and SMOTE taught me the hard way.

Aug 15 · 8 min read · godfident-data.hashnode.dev  
Read the article

-   [#machinelearning](https://hashnode.com/tag/machinelearning)
-   [#datascience](https://hashnode.com/tag/datascience)
-   [#imbalanceddata](https://hashnode.com/tag/imbalanceddata)
-   [#frauddetection](https://hashnode.com/tag/frauddetection)
-   [#precisionrecall](https://hashnode.com/tag/precisionrecall)
-   [#smote](https://hashnode.com/tag/smote)
-   [#logisticregression](https://hashnode.com/tag/logisticregression)
-   [#learninginpublic](https://hashnode.com/tag/learninginpublic)
-   [#mlmetrics](https://hashnode.com/tag/mlmetrics)
-   [#creditwiseai](https://hashnode.com/tag/creditwiseai)

## Written by

[Bahati Brenda Kizito](https://hashnode.com/@Lioness256)

@Lioness256

  
[View profile](https://hashnode.com/@Lioness256)

Documenting my journey into data and AI

## Responses

Markdown supported · ⌘↵ to post

No responses yet.