Scribd
Upload
68 views21 pages

Wine Quality Classification Using Weka

Uploaded by

baburaogapte30
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
68 views21 pages

Wine Quality Classification Using Weka

Uploaded by

baburaogapte30
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/ 21

Wine Quality

Classification
Using Weka
Darshan Pathak 123M1H041

Supervised By:
Prof. Bali Khurana
01
Classification
What is Classification
Classification in Machine Learning aims to determine which category an
observation by understanding the relationship between the dependent
and independent variables.

Classification algorithm can learn


to predict whether a given email is
spam or not span
Working of Classification Algorithms
Classification algorithms sort data into predefined categories based on
patterns they learn from labeled examples. They use features (data
attributes) to create a model during training, and then apply this model
to predict the classes of new, unseen data.
Types of Classification

Binary Classification Multi-Class Classification

Multi-Label Classification Imbalanced Classification


02
Case Study
Let’s learn with help of

WineQuality Case Study


Using Weka Software
What is WEKA Software

• Waikato E nvironment for Knowledge A nalysis


• Collection of machine learning algorithms and data processing
tools implemented in Java
• Used for the process of experimental data mining
• Preparation of input data
• Statistical evaluation of learning schemes
• Visualization of input data and the result
1. Install Weka
2. Load Data
Wine Quality Dataset
3. Visualize
4. Select J48 Classifier
5. Evaluate Result
Summary
Classification Corretly
Classified
Kappa statistic 0.3952
61.4% Classifier - J48
Mean absolute
error
0.1359 982 Instances
● Decision tree classifier
● Recursively splits based on
Root mean
squared error 0.3332 attribute
● Select features to create
decision nodes and
Relative
absolute error
63.3475 % Incorrectly branches.
● Emphasizes information
Classified gain.
Root relative
squared error 101.8207 % 38.5%
● Effective for categorical data.

Total Number of 617 Instances


Instances
1599
Proportion of actual positive instances correctly
TP Rate identified.

Proportion of actual negative instances incorrectly


FP Rate classified as positive.

The accuracy of positive predictions, representing the


Precision ratio of true positives to the total predicted positives.

The model's ability to correctly identify all actual


Evaluation Recall positive instances.
Metrics
Combines precision and recall into a single metric,
F-Measure balancing both aspects of classification performance.

Matthews Correlation Coefficient considers true &


MCC false +tives/-tives to assess classification quality

Measures the model's ability to distinguish between


ROC Area classes

Quantifies the model's precision-recall trade-off


PRC Area
Detailed Accuracy By Class
F-
TP Rate FP Rate Precision Recall MCC ROC Area PRC Area Class
Measure

0.1 0.006 0.091 0.1 0.095 0.089 0.541 0.026 1

0.113 0.032 0.107 0.113 0.11 0.079 0.526 0.047 2

0.711 0.237 0.689 0.711 0.7 0.472 0.76 0.652 3

0.614 0.259 0.612 0.614 0.613 0.355 0.706 0.579 4

0.497 0.056 0.559 0.497 0.527 0.465 0.789 0.418 5

0 0.008 0 0 0 -0.009 0.611 0.022 6

Weighted
0.614 0.213 0.611 0.614 0.612 0.403 0.732 0.563
Avg.
Confusion Matrix
a b c d e f

a 1 3 2 3 1 0 a=1

b 4 6 24 16 3 0 b=2

c 2 24 484 154 17 0 c=3

d 3 19 169 392 50 5 d=4

e 1 4 23 65 99 7 e=5

f 0 0 0 11 7 0 f=6

A confusion matrix is a compact table summarizing the performance of a classification


model, detailing true positive, true negative, false positive, and false negative predictions
for each class, aiding in model evaluation and error analysis.
03
Conclusion
This presentation on wine quality classification using the
J48 algorithm in Weka highlights the significance of
classification in machine learning, specifically exploring the
Weka software and its application to the Wine Quality
dataset. The detailed accuracy metrics and confusion
matrix provide valuable insights into model performance,
aiding in informed decision-making using classification in
Data Mining
Thank You

You might also like