Differences

This shows you the differences between two versions of the page.

--- week06 [2021/07/21 08:37] – suhaw
+++ week06 [2021/07/22 06:29] (current) – [Ghorra Paul] suhaw
@@ Line 1: / Line 1: @@
 ====== Week 005 ======
+===== Thursday 22 July 2021 =====
+==== elearning - Introduction to Business Analytics ====
+==== John David Ariansen and Madecraft ====
+Data sources:
+  * Internal
+    * Sales
+      * Breakdown to: demographics, geographic, time, product category, average sale
+    * Cost
+      * Fixed
+      * Variable
+    * Marketing
+    * Psychometric
+  * External
+    * Market Reports
+      * Market players
+      * Size of market
+      * High-level data points
+    * Market Research
+      * Behavourial patterns
+      * Buyer personas
+Data governance: ensure quality of data assets
+Causes of data quality issues:
+  * Data collection: Manual, lack of index field
+  * System integration (different data format across systems)
+  * Missing data
+Tools/Applications:
+    * Sales: Square, Stripe, Clover
+    * Cost: Quickbooks
+    * Marketing: Google Analytics, Salesforce
+    * Psychometric: Google Forms, Qualtrics, Social Media data, focus groups
+Data Management tools:
+  * MS Excel
+  * Tableau
+  * PowerBI
+  * MS SQL Server
+==== Ghorra Paul ====
+BCG Report "Putting AI to Work", Sept 2017 https://www.bcg.com/publications/2017/technology-digital-strategy-putting-artificial-intelligence-work
+Shipping container routing: https://youtu.be/QQmZiQfeVNw
+Key takeaways:
+  * Gathering relevant data and preparing it for analysis is often the longest step of analysis
+  * Make it visible - Just visualising existing data can already unlock tremendous value
+  * Simulation: A digital twin can expedite learning with risk free experimentation
+  * ML: Input data determines strength of the algorithm - it cannot predict outcomes it hasn't seen
+  * 3 step framework - Decision variables, Objective, Constraint, help to frame optimisation problem
@@ Line 9: / Line 71: @@
 === AI ===
+  * Symbolic Reasoning/Planned AI
+    * Abstract problems, but know steps
+  * Machine Learning
+    * Look for patterns
+  * Artificial neural networks
+== Approaches ==
+  * Match patterns
+  * Data vs reasoning
+  * Unsupervised learning
+    * Machines creates categories based on similarities it detects, may be different from human categories
+  * Deep learning
+  * Clustering
+  * Backpropagation
+    * Allows later feedback to go back and adjust earlier rules/algorithms, error corrections
+  * Regression
 === IoT ===