ML Breath

[C3.ai] Model Building End-To-End

[Uber] L1 & L2 Regularization

[Google] Regression Model Metrics

[Amazon] Euclidean and Manhattan Distances

[Capital One] Feature Selection

[Amazon] Underfitting and Overfitting

[Google] Model Evaluation

[Uber] Explain AUC

[Amazon] The Curse of Dimensionality

[Capital One] Handling Missing Values

[Amazon] Hyperparameter Tuning

[Salesforce] Outliers in Modeling

[PayPal] Imbalanced Labels

[Google] Multicollinearity

[Amazon] Cross-Validation

[Amazon] Variance-Bias Trade-Off

[Amazon] Model Ensembling

[Amazon] Handling Categorical Variables

[Meta] Feature Engineering

[Microsoft] Text Features

ML Algorithm

[Microsoft] Finding Optimal K in K-Means

[Amazon] GBTs Hyperparameters

[Amazon] GBTs Pseudocode

[Apple] KNN Pseudocode

[Microsoft] K-Means

[Microsoft] Principal Component Analysis

[Salesforce] K-Means on Scaled and Unscaled Data

[Intuit] Variance & Bias in Trees

[Amazon] Bagging vs Boosting

[PayPal] Random Forest vs Logistic Regression

[Meta] Decision Tree Pseudocode

[Intuit] Logistic Regression Pseudocode

[Google] How do you evaluate regression modeling?

[Meta] How Is the random forest model trained?

Deep Learning

[PayPal] Activation Functions

[Microsoft] Gradient Descent

[Microsoft] Backpropagation

[Google] RNN vs CNN

[Amazon] Neural Network Overfitting

[Intuit] Encoder-Decoder

[Google] Tuning Dense Neural Networks

[Google] Normalization

[Google] Optimizers

ML Production

[Intuit] Kubernetes in Model Deployment

[Capital One] Model Failure in Production

[Intuit] Model Deployment on AWS

[Google] Online vs Offline Model Performance

[PayPal] Prediction Point

[Capital One] Model Deployment Challenges

[Intuit] Model Orchestration

Machine Learning

<p><strong>Last Updated</strong><em><strong>:</strong>&nbsp;August 2021</em></p>

<h2 class="wp-block-heading"><strong>Problem</strong></h2>

<p>What is AUC? How is it helpful when labels are imbalanced?</p>

<hr>

<h2 class="wp-block-heading"><strong>Solution</strong></h2>

<p>Suppose there are a total of 100 observations – 99 goods and 1 bad. Your classification model predicts all 100 observations as good, resulting in 99% accuracy. The model is great at correctly classifying goods with a 100% true positive rate, but horrendous at classifying the bad with a 100% false positive rate.</p>

Last Updated : &nbsp;August 2021 Problem What is AUC? How is it helpful when labels are imbalanced? Solution Suppose there are a total of 100 observations…

DataInterview

[Uber] Explain AUC

Problem#

Solution#

Unlock the full lesson

Problem

Solution