Predictive Modelling and Machine Learning

Synopsis:

This course introduces the principles, theories and concepts of statistics and data modelling. Students will learn how to construct and interpret graphical presentations of data, conduct appropriate statistical tests, use the appropriate techniques in data modelling and interpret the results generated. The students will be applying these statistics and data modelling techniques in practical projects, through which they will learn how to develop real world analytics solutions using RapidMiner.

Prerequisite: Nil

Topics

2020 March

Materials on Google Class room

Schedule

Venue: UE0002

Assessment:

1. Lab Practical 15%
2. Group Assignment 35%
3. Written Exam 50%

Instructors:

1. Ai Huey LIM lim_ai_huey@nyp.edu.sg
2. Joanne FOO joanne_foo@nyp.edu.sg

Extra Materials

Tools and Softwares

1. Microsoft Excel
2. Tableau Desktop Student license will be issued during the lab session. To activate Tableau on PC behind the proxy firewall, look here .
3. RapidMiner Studio Trial license allows 1000 rows of test record. To register RapidMiner on PC behind the proxy firewall
   1. Launch rapidminer studio
   2. When the application prompts for license, choose “manually enter the license”
   3. The license key can be applied and retrieved from "My Rapidminer".
   4. Copy paste the key from the link above into the license key window in RapidMiner Studio

References:

1. Statistics for Managers: Using Microsoft Excel (6th ed.), David M. Levine et al. (2011), OT, B.
2. Foundations of Predictive Analytics, James Wu & Stephen Coggeshall (2012), OT, B.
3. Data Mining: Concepts and Techniques, Jiawei Han, Micheline Kamber, and Jian Pei, (2011), OT, B.

Old materials

2017 October

1. Statistics Essentials (Part 1 || Part 2)
   1. Descriptive Statistics
      1. Types of data
      2. Central Tendency
   2. Data Visualisation
      1. Types of Charts
      2. Dashboard Interaction
      3. Animation
   3. Inferential Statistics
      1. Univariate Statistical Techniques
      2. Multivariate Statistical Techniques
      3. Parametric and non-parametric tests
2. Predictive Modelling (Part 1 || Part 2 || Part 3)
   1. CRISP-DM model
   2. Supervised Learning
      1. Classification
      2. Decision Tree
      3. Linear Regression / Logistic Regression
   3. Unsupervised Learning
      1. Association Rule Mining
      2. Types of Clustering
   4. Applications of Predictive Models in Domains
3. Scala, Spark and Machine Learning
   1. Scala
      1. slides
      2. exercises
      3. source codes
   2. Spark Machine Learning
      1. slide 1
      2. exercise 1
      3. slide 2
      4. exercise 2
   3. Exercise and discussion

Assessment:

1. Practical assignment (individual, 50%)
2. Group research report and presentation (2-3 in a group, 50%)
   • Suggested topics. But you are not limited to these and you are encouraged to find out other topics and applications of predictive modelling.
   • Pick an application of Machine Learning, e.g. sentiment analysis, facial recognition, object identification, search-based regression test prioritization which can be solved by using at least two different approaches (two different ML approaches).
   • Study and research into papers and tools on related to these approaches in depth.
   • Compare these approaches in the context of the application of study.
   • Write up a summary on the problem, the solutions and the comparison.
   • A 10 minutes presentation is required.
   • Deadline 15 Mar 2019
   • Email your presentation slides and report to keith_fwa@nyp.edu.sg