IMW - Classification Modelling
Venue: Indian Institute of Technology Bombay, Mumbai
Dates: 20 Feb 2023 to 25 Feb 2023
Convener(s)
| Name: | Prof. Sanjeev Sabnis | Prof. Radhendushka Srivastava |
| Mailing Address: | IIT Bombay (Mathematics) | IIT Bombay (Mathematics) |
| Email: | svs at iitb.ac.in | rsrivastava at iitb.ac.in |
Please Note:
- NCM Participants are requested to register in NCM website followed by registration in CEP website: http://www.cep.iitb.ac.in (through Google Chrome browser).
- Shortlisted NCM participants will have to pay nominal course fee of Rs 1000 + 18% (GST) = Rs 1180
- Last Date of receiving online application from participants is 31 Dec 2022 for both NCM and CEP website
Note: Lab sessions will be held in Python. Participants are required to know the basics of Python.
Description
The word ‘classification’ in the context of statistical modelling refers to a classification of new observations into relevant classes using a statistical decision rule that is built using training data pertaining to a particular phenomenon or a field.
This classification exercise is pervasive across fields such as medicine (for example, classifying individuals having COVID or not having COVID into classes such as ‘severe symptoms’, ‘non-severe symptoms’, ‘absence of symptoms’), various manufacturing industries (classifying newly manufactured items into ‘defective’ and ‘non-defective’ classes), banking (classifying clients into classes such as ‘fraudulent’ and ‘non-fraudulent’), social sciences and law and many more fields.
The classification methods that will be covered in this workshop include (i) logistic regression, (ii) linear and quadratic discriminant analysis, (iii) naïve Bayes, (iv) K- nearest neighbours, (v) decision trees, (vi) random forests, (vii) support vector machines. Each of these classification methods will be demonstrated using real and simulated data with the help of open-source software R/python. In the machine learning parlance, these classification methods are also referred to as supervised learning methods.
Some unsupervised learning techniques (mainly, clustering methods) will also be covered in the workshop. The K-means clustering and hierarchical clustering algorithms will be demonstrated through real data.