whitfordr/httlads

How to Think Like a Data Scientist

How to Think Like a Data Scientist

This is a project based course to get you to think like a data scientist, teach you some of the tools of a data scientist and generally improve your problem solving skills.

Course Outputs Learning Objectives

• Articulate the data processing
• Extract data using SQL
• Gather data from the Internet; combine data from sources
• Clean the data
• Handle missing data/finding outliers/fixing data
• Normalize the data
• Visualize the data
• Translate questions to analysis
• Analyze data
• Single variable regression, logistic regression
• Market basket analysis
• Cohort analysis
• Sentiment analysis, exposure to Bayes
• Time series
• Geographic analysis
• Simulations, Monte Carlo
• Run basic statistical tests, set up null hypotheses

Course Outline

[Week 1-2] -- Module 1

• Spreadsheets. Ex: Environmental studies, weather, happiness * data on countries

  • Descriptive statistics
  • Charts: Scatter plots, box plots, line plots, histograms

• Using functions, including vlookup

• Optimization using Solver

[Weeks 3-4] -- Module 2

Review of Frequency Analysis for Textual Data. Data from United Nation dataset.

• Brainstorming/Generating questions: What were hot topics? What topics are common?
• Using Word Cloud & textatistic library in Python
• Using Web APIs
• Join two datasets

End-of-Module Exercise: Pick your favorite out-of-copyright data repository, and analyze character references. Project Gutenberg.

• Module Intro (1 day)

  • Loading files
  • This will be good review for python usage, e.g. reading files, working with dictionaries, etc.
  • Give an example of a question on something simple: what were important topics discussed in 1970?
  • From this dataset, ask students what other questions would they want to explore, jot down on the board. Add seeded questions to the board (from facilitator guide).
  • Segue into answering student-driven questions using Pandas; framing Pandas as a way to address pain felt earlier

• Pandas Intro

  • Prework: Watch Tutorials to get intro to Pandas: https://www.youtube.com/watch?v=5JnMutdy6Fw
  • Data Exploration Exercise #1 (Do this, what happened?, modify it, repeat):
  • Load data into data frame
  • Use filtering to see only the data from 1970)
  • see only data from 1970 from U.S.
  • Group By Year
  • Group By Country
  • Answer one of the questions you generated above, using some of the things you just learned/explored.

• Data Exploration #2: Quantify Text Data using

  • Use wordcloud to create a visual, introduce that it’s minor data cleaning (removes ‘and’, ‘the’)
  • Use textatistic to assign a number to each row (add a column to your dataset)
  • Instructor-led debrief to review solution(s) and relay tips
  • Plot the complexity calculated

• Screen Scrape CIA data

  • CIA data is available online but not very convenient. The CSV file we have is from 2006 so is quite outdated. We would like to update this data.
  • Tutorial using requests python module
  • https://towardsdatascience.com/data-analytics-with-python-by-web-scraping-illustration-with-cia-world-factbook-abbdaa687a84

• Data Exploration #3:

  • Add continent, region, GDP, life expectancy to our dataset
  • Create a simple visualization using newly added data (e.g. showing each country on the map with life expectancy)
  • Save the dataset by exporting to CSV so that you don't have to pull the data again

[Weeks 5-6] Module 3

• Review of Frequency Analysis for Numeric Data. A numeric-based exercise from say DataUSA, pop culture (movie gross?)
  • Brainstorming/Generating questions: What were hot topics? What topics are common?
  • Reading CSV/JSON
  • Handle missing data
  • Normalize the data
  • Correlation
  • Regression or classification
  • Exercise: TBD

[Week 7-8] Midterm Project Weeks

Work in pairs, choose your dataset In-class work on projects -- practice applying concepts from modules above

[Week 8-9] Module 4

Predictive -- simple recommender system. Ex: Insta-cart shopping basket recommendation; MovieLens data; beer & diapers story; movies/music/books.

• Correlation
• Pull in Heat maps
• SQL basics
• Bias

[Week 10-11] Module 5

• Time Series/Historical Data Analysis. Ex: Hamburger prices
  • Currency converting
  • Histograms
  • Sampling
  • Hypothesis testing

[Week 12, Holiday Break, TBD & movable] Fun Model + Choosing Visualizations

• Simulation -- Monte Hall -- 3 Doors, Prize
• Choosing between chart types for reports
• Dice or Roulette simulation
• Extra alternatives: Cohort Analysis

[Week 13-14] Final Project & Presentation Week

Work in pairs, choose your dataset Presentations (possibly a poster fair) & feedback