Data Science Principles from Harvard Business School Online

Module 1: Data 101    

Case Studies

• Flu Detection

Takeaways

• Explain why data collection is important

• Identify factors that may affect data quality

• Recognize that not all data is numerical

• Explain how the organization of data can affect the information you are able to extract from it

Key Exercises

• List sources of data

• Discuss what can be done with data

• Categorize data by various factors

• Determine whether data is high-quality or not

Module 2: Predictions and Recommendations 

Case Studies

• Predicting Sepsis

Takeaways

• Understand the basic structure of a predictive algorithm

• Identify where human decisions shape predictive systems

• Evaluate the success of a predictive system

Key Exercise

• Examine how weather forecasts work

• Use data to create a prediction

• Sort types of training data

• Simulate a predictive system

Module 3: Cause and Effect   

Key Exercise

• The Google Tax

Takeaways

• Explain why it is important to establish causal relationships

• Identify barriers to establishing causal relationships in a variety of settings

• Identify why randomization can help establish a causal relationship but also create other problems

Key Exercises

• Classify relationships based on correlation or causation

• Examine the relationship between variables

• Identify potential common causes for correlated events

Module 4: Data Governance and Privacy         

Case Studies

• Privacy and Facial Recognition

Takeaways

• Explain why data privacy is important

• Describe what can constitute a violation of privacy

• Critique existing privacy policies

• Create a set of ethical tenets to guide data work at their own organizations

Key Exercises

• Formulate data privacy guidelines

• Discuss the risks of data re-identification

• Evaluate existing data privacy policies for ethics

Module 5: Beyond the Spreadsheet

Case Studies 

• Burning Glass and Text Data

Takeaways

• Identify sources of non-numerical data

• Explain why it would be useful to use non-numerical data

• Describe the differences in approach for supervised and unsupervised learning

• Identify use cases for neural networks

Key Exercises

• Perform a sentiment analysis

• Determine what types of data an algorithm cannot read

• Examine how computers intake visual and audio data

• Experiment with facial recognition

Module 6: Data Science Ecosystems  

Case Studies

• Harvard Link

Takeaways

• Explain the importance of data transformation and wrangling

• List the common technologies used within data science ecosystems

• Describe the connection between data science tasks, software tools, and hardware tools

• Identify potential sources of bottlenecks in the data science process

Key Exercises

• Identify and order the lifecycle of data

• Define what "the cloud" is

• Estimate the size of various data streams

Module 7: The Road Ahead

Case Studies 

• Healthcare Prioritization

Takeaways

• Recognize a problem that an algorithm might be able to solve

• Recognize the challenges created by using data science tools in ways outside their intended use

• Identify steps within the data science process that need auditing

Key Exercises

• Choose types of data to ingest into an algorithm

• Evaluate the risks of solely using an algorithm to make decisions

• Discuss how algorithms can reinforce biases

• Create a set of guidelines to evaluate projects