We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 15
Modeling with
Instances Prof. Sonia F. Panesar
Department of Computer Sci. & Eng. Babaria Institute of Technology Classificatio n Classificatio n • Classification is a form of supervised machine learning • The classification algorithm learns from labeled data. Data labels make it easier for your models to make decisions based on the logic rules you’ve defined Classificatio n • Instance-based learning classifiers are supervised • lazy learners — they have no training phase, and they simply memorize training data, in- memory, to predict classifications for new data points Instance-based classifiers • k-nearest neighbor (kNN) • Self-organizing maps • Locally weighted learning limitations • These classifiers are not well-suited for – Noisy data (data with unexplainable random variation) – Datasets with unimportant or irrelevant features – Datasets with missing value Clustering Vs Classification Paramenter CLASSIFICATION CLUSTERING
Type used for supervised learning used for unsupervised learning
grouping the instances based on
process of classifying the input instances their similarity without the help of Basic based on their corresponding class labels class labels
it has labels so there is need of training and
testing dataset for verifying the model there is no need of training and Need testing dataset created
less complex as compared to
Complexity more complex as compared to clustering classification
k-means clustering algorithm, Fuzzy
Example Logistic regression, Naive Bayes classifier, c-means clustering algorithm, Algorithms Support vector machines etc. Gaussian (EM) clustering algorithm etc. Nearest Neighbour Analysis • Nearest neighbor methods work by taking an observation’s attribute value and then locating another observation whose attribute value is numerically nearest to it Average Nearest Neighbour Analysis • Average nearest neighbor algorithms are basic yet powerful classification algorithms. They’re useful for finding and classifying observations that are most similar on average • used in pattern recognition, in chemical and biological structural analysis, and in spatial data modeling. • The purpose of using an average nearest neighbor algorithm is to classify observations based on the average of the arithmetic distances between them. K Nearest Neighbour Analysis How it Works Usefulness & Limitation • Useful for multi-label learning — supervised learning • It takes a lot longer than other classification methods to classify a sample • performance depends on the distance function and on the of neighborhood parameter k. value the Applications • website categorization, web page ranking • customer relationship management (CRM) Real-World Problems with Nearest Neighbor Algorithms • used in retail to detect patterns in credit card usage. • use for visual pattern recognition to scan and detect hidden packages in the bottom bin of a shopping cart at checkout. If an object is detected that is an exact match for an object listed in the database, the price of the spotted product could even automatically be added to the customer’s bill. Reference s• https://analyticsindiamag.com/7-types - classification-algorithms/ • https://matthew- brett.github.io/dsfe/chapters/ 09/Nearest_Neighb ors • https://www.analyticsvidhya.com/blog/2018/08 / k-nearest-neighbor-introduction-regression- python/ • https://www.analyticsvidhya.com/blog/2018/03 / introduction-k-neighbours-algorithm-clustering /
DATA MINING and MACHINE LEARNING. CLASSIFICATION PREDICTIVE TECHNIQUES: SUPPORT VECTOR MACHINE, LOGISTIC REGRESSION, DISCRIMINANT ANALYSIS and DECISION TREES: Examples with MATLAB