Scribd
Upload
6 views5 pages

14.1 Data Preprocessing Class Imbalance and AUC Curve

The lecture covered essential data preprocessing steps including data cleansing, feature engineering, and handling missing data. It also addressed class imbalance through techniques like stratified k-fold cross-validation and the use of micro and macro-averaged recall/precision. Finally, the ROC-AUC curve was discussed, highlighting the significance of AUC values in evaluating model performance.

Uploaded by

dev3421667
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
6 views5 pages

14.1 Data Preprocessing Class Imbalance and AUC Curve

The lecture covered essential data preprocessing steps including data cleansing, feature engineering, and handling missing data. It also addressed class imbalance through techniques like stratified k-fold cross-validation and the use of micro and macro-averaged recall/precision. Finally, the ROC-AUC curve was discussed, highlighting the significance of AUC values in evaluating model performance.

Uploaded by

dev3421667
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 5

Recap of Lecture on Data

Preprocessing, Class imbalance


and ROC –AUC curve
Data Preprocessing steps
• Data Cleansing:
• Tackle missing data
• For text applications, remove stop words, convert all verbs to root forms
• Remove inconsistencies (rows with same attribute values but different labels)
• Feature engineering – Converting raw data to a usable form for ML.
Ex
• Feature rejection (drop unrelated features) and feature selection
• Feature Transformation: transform raw data with PCA and reduce
dimensionality by using only important components
• Feature extraction: Generate new features by combining existing features
Tackling missing data
• Missing values
• Delete rows with missing data
• Delete columns if it has multiple empty entries
• Assign all possible values if categorical
• Assign average value of column, if values are continuous. Assign median value
of column if categorical
• Decision relative mean/median: Use only rows with same label as that of row
with missing value
• Closest Assign feature value with closest matched row with same label or
with centroid of closest matched cluster with same label
Handling class imbalance
• Use stratified k fold cross validation – each fold has same ratio of two
or more classes
• Use micro-averaged recall / precision if you want that class with
greater representation should be given more priority (Microaverage
gives equal priority to each instance)
• Use macro-averaged recall / precision if you want that class with less
presentation also gets equal priority as other classes. (Macroaverage
gives equal priority to all classes)
ROC – AUC Curve
Receiver Operating Characteristics - Area Under the Curve

AUC of Curve 1: 0.92 (best)


AUC of curve 2: 0.75
AUC of Curve 3: 0.5
Curve 1
Curve 2
Sensitivity or Recall

Curve 3

Specificity

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy