Data Science Project Planning

Importance of business problem definition. Consequences of poor business problem definition; Advantages of well defined problems. Tasks involved in business problem definition.

Identifying types of business problems ; examples of business problems

This lecture will describe how to identify project stakeholders and gather their inputs

This lecture will describe document the inputs obtained from stakeholder and assess a stakeholder's influence and interest in the project

Provides pointers to the sources of information about the previous work done to solve the business problem. Also explains why one should review the previous work.

This lecture is about creating a Business Problem Statement and getting the Stakeholder Buy-in for the same.

Introduction to various phases of CRISP-DM, a popular data science project life cycle based on scientific method for solving problems

A brief mention of steps involved in formulating a data science problem

Conceptual level understanding of Classification Problem

Conceptual level understanding of Regression Problem

Conceptual level understanding of Clustering Problem

Conceptual Understanding of Anomaly Detection Problem

Conceptual Understanding of Association Problem

Conceptual Understanding of Recommendation problem

All the six data science problem types discussed in the previous lecturesummarized in a tabular form with additional examples and also a single page visual recap.

A brief discussion on three areas which data science project goals should focus on  viz; model development, establishing data flow pipeline and documentation

This overview lecture will briefly mention

a) the metrics used to evaluate model quality, their deployment and monitoring. 

b)  the metrics used for evaluating the efficiency and effectiveness for data flow pipeline

c) checkpoints to evaluate documentation quality

Definitions of  Accuracy, Precision, Recall, F1 Score with examples

Definitions of Metrics - Precision, Recall, F1 Score; Accuracy Paradox;

Metrics discussed - Root Mean Square Error (RMSE), Coefficient of Determination (R-Squared)

Dunn Index & Silhouette Coefficient

Rand Index & Jaccard Index

Support, Confidence, Lift

Metrics to measure Prediction Error - Mean Absolute Error(MAE), Root Mean Square Error (RMSE),

Metric to measure Relevance -Mean Average Precision(MAP)

Metrics to measure Diversity, Coverage and Serendipity

Multivariate Testing; Effect Size; Statistical Power

Population Stability Index (PSI); Resource Consumption; Cost-Effectiveness

Availability; Latency; Throughput; Integrity; Scalability; Security; Privacy

Project Documents; Main Contents; Structure; Writing Style; Visuals

Enumerates the factors that should be assessed while setting goals and planning a data science project

Skills needed, Team Roles & Team Attributes

Assessment of  available Data, Knowledge and Computing resources

Different kinds of requirements the project may have to satisfy; assumptions underlying the project plan; constraints the project may have to operate.

General project risks; data related risks; risk assessment criteria; mitigation and contingency planning

Need for a glossary of terminologies

Factors to consider while conducting cost/benefit analysis

Overview of key deliverables, activities and team roles for each phase in CRISP-DM lifecycle

Discusses the attributes of  an effective schedule

A brief overview of how Emerging methods extend CRISP-DM through Agile approach

Overview of Key components of Team Data Science Process

Overview of Agile Data Science Manifesto and the main topics discussed in the book Agile Data Science 2.0 by Russell Jurney

A quick summary of main contents of this course.

List of questions from both business and data science problem perspective  that needs to satisfactorily answered by the project plan before proceeding further

Important points to bear in mind while planning a data science project.