• Role of Data Science and DS Professionals

Data-driven decision making in the business world
The data professional’s ecosystem
Data science as a strategic asset for businesses across industries
The roles of a data analyst and data scientist
Survey of storytelling with data

• Data Science Solutions to Business Problems

Key business operational concepts
Typical business problems
Role of data to solve business problems
The process of data analytics
Data science modeling - supervised and unsupervised
Anatomy of a business case with data analysis for marketing

• Data Science Solutions to Business Problems

General business problems in the healthcare and pharmaceutical
industries
Characteristics of data science projects in these industries
Anatomy of a business case wherein data analytics offered a solution
to the problem

• Conceptualization of a Data Analytics Problem

Planning process for a data analytics project
Understand the business problem
Data available for the project
Review of dataset(s) to see what knowledge and insights can be extracted
Establish project objective(s)
Preparing data for analysis
Determine tools and methods
Envision the approach
Identify deliverables
Tell the story to stakeholders

• Simple Linear Regression

Overview of Machine Learning
Residuals, RSS, and the Coefficient of Determination.
Assumptions of Simple Linear Regression.
Model coefficients:
■ Interpretation
■ Standard Errors

• Multiple Linear Regression

Assumptions of Multiple (General) Linear Regression
Coefficients of Continuous Features
Multicollinearity and Overfitting
Dummifying Categorical features:
■ Coefficient Interpretation
Case Study

• Penalized Linear Regression

The Bias-Variance tradeoff
Euclidean (L2) vs Manhattan (L1) Distances
Ridge, Lasso, and ElasticNet Penalties
Hyperparameter grid search and the model coefficient shrinkages

• Logistic Regression and Gradient Descent

The failure of Linear Regression Models for Classification
Log-odds and Sigmoid function
The concept of Maximum Likelihood Estimation
Negative Log-Likelihood as Loss
Gradient Descent and Stochastic Gradient Descent

• Model Selection

Model assessment in practice:
■ Grid search
■ Train-test split
Model comparison:
■ Tuning
■ Various model types
Thoughts on Feature Engineering:
■ Different features make different models
Case study