Chapters 2-7 (Chapter 5 Decide on Unit Analysis (5.1 What is a Unit of…
Chapter 5 Decide on Unit Analysis
5.1 What is a Unit of Analysis?
The main focus for unit of analysis is that it is truly the who what when where why and how of the data. The what is what the actual data is about the who could be who it is about.
5.2 How to Determine Unit of Analysis
One way to identify the appropriate unit of analysis is to first think about what the prediction target is, which will then oftentimes supply an obvious choice.
Chapter 6 Define Prediction Target
6.1 What is a Prediction Target
The main thing about the prediction target is that it is trying to predict the future. But it is always near impossible to predict what will happen next.
6.2 How is the Target Important for ML
The reason we need a target is because it is important to figure out what connect with what. It is near impossible for people even machines to understand why something occurs based on another variable.
Chapter 7 Success, Risk, and Continuation
7.1 Identity Success Criteria
This is important for understanding how successful data might be. We use to see who will use the model, to see who is on board especially management, and how much the model will be used or successful.
7.2 Foresee Risks
The first and most important to thing to understand is that risks are hard to detect.
Analytics is used not only for core capabilities, but also in related functions like Marketing and Human Resources.
Top management team recognizes and drives use of analytics and innovative measures of success. Fact-based decision-making is part of the organizational culture. Copious data and analytics is performed, and results ares hared both internally and with partners and customers.
Chapter 2 Automating Machine Learning
2.1 What is it?
It is basically any machine learning system that automates the repetitive date that goes into machine learning. The first step in Machine Learning is Exploratory data analysis. This is the process of examining the descriptive statistics for all features as well as their relationship with the target. Second step feature engineering, third is selecting the algorithm, then lastly is model diagnostics.
2.2 What it's not
Technically speaking it is not all automatic. It is actually more conceptual then math based. ML is good news for data scientists because it frees them from manually testing out all the latest algorithms, the vast majority of which will not improve the performance of their work.L
2.3 Available Tools and Platforms
In general there are two types of tools for AutoML. "Leads, these are very conceptual. There are many tools available, such as Wise.io which is owned by General Electric, but there are many other tools.
2.4 Eight Criteria
The 8 Criteria are: Accuracy, Productivity, Ease of use, Understanding and Learning, Resource Availability, Process Transparency, Generalizing, and lastly Recoommend Actions.
Chapter 3 Specify Business Problem
3.1 Why Start with a Business Problem?
Rather than looking at these issues as problems it is important to use the word opportunity as well. We need to find out what the problem is and how are we able to solve this problem in the current market.
3.2 Problem Statements
This part also discusses many questions that need to be answered. Are the problems that are being asked, important to the business. When looking at figure 3.2 in the book, might provide an improved project statement by outlining how much the hedge fund is willing to invest annually, what kinds of interest rates may be received.
Chapter 4 Aquire Subject Matter Expertise
4.1 The Importance of Subject Matter
Beyond knowledge of what the features mean, another important task for which we need a subject matter expert is setting realistic expectations for model performance.
lt is also important to expect the subject matter expert to suggest ideas for data collection, that is, to know where relevant data resides, including any external data.