Comprehensive Data Science with Python

Software Requirements

• Anaconda Python 3.6 or later
• Spyder IDE and Jupyter notebook (Comes with Anaconda)

Outline

An Accelerated Introduction and Overview to Python for Data Science Foundations

• Introduction to course and computing environment
• Up and running with Jupyter notebooks
• Fundamental Python types: String literals, numeric, Boolean, and dates
• Understanding Python ‘variables’ (reference assignment)
• Slicing syntax
• Fundamental collections: tuples, lists, dictionaries, and sets
• Control flow iteration in Python (if/then, for, while, list comprehension)
• Writing your own functions
• Handling exceptions

Matrix Computing with NumPy

• Introduction to the ndarray
• Dtypes in NumPy
• NumPy operations, uFuncs
• Broadcasting
• Missing data in NumPy (masked array)
• Random number generation

Managing, Exploring, and Cleaning Data with Pandas

• Fundamental Pandas: Series and DataFrames
• Exploring objects with attributes/methods
• Importing data from different structured sources
• Basic DataFrame summaries
• Creating new variables (columns)
• Scaling and standardizing data elements
• Discretizing continuous data
• Mapping categorical data to new values
• Establishing dummy codes (one hot encoding)
• Filtering rows and selecting columns
• Managing the indices
• Identifying duplicate rows
• Quantifying and managing missing data
• Combining datasets
• Merging datasets
• Transposing datasets
• Changing data from long to wide formats and back

Exploratory Data Analysis with Pandas (including visualization with Seaborn)

• Univariate Statistical Summaries and Detecting Outliers, visually with graphical approaches and numerically.
• Multivariate Statistical Summaries and Outlier Detection, visually with graphical approaches and numerically.
• Groupwise calculations
• Pivot Table type operations to aggregate by group
• Pandas DataFrame plotting methods

Data Pseudo-Coding Process, Extension to Data-Centric Problems

• Identifying data verbs
• Answering a question using a well-formatted analytic dataframe
• Understanding the unit of analysis
• Identifying the unit of analysis for a given question – is my dataframe organized this way?
• Leveraging normalized data to create the analytic dataframe through combinations of data verbs
• Identify the question and unit of analysis
• Define the desired analytic dataframe
• Examine the normalized source data
• Create data pseudo-code to map source data to the final analytic dataframe
• Implement with Python

Focus on Graphics with Python: Seaborn, Matplotlib, and Plotly

• Using seaborn for 1 and 2 variable summaries
• Advanced statistical plots with Seaborn
• Controlling plot details through Seaborn
• Making graphs interactive with Plotly
• Introduction to Matplotlib for full control of parameters

Overview of Descriptive versus Inferential Analytics

• Identifying the null hypothesis
• P-value interpretation
• The idea of statistical power and type 1/2 errors

Implementing Inferential Statistics in Python

• Analyzing an A/B randomized test:
• T-tests/ANOVA
• Chi-square tests
• Correlation methods

Multivariate Models: Linear Regression

• Estimating the mean
• Identifying p-values of interest
• Adding a categorical predictor and the link to t-tests
• Nonlinear trends: Polynomial regression and spline modeling
• Interaction terms
• Confounding
• Model building approaches (choosing the best model)
• Scoring new data from the model (making predictions)

Multivariate Models: Logistic Regression

• GLMs and the link function
• Understanding the logit function
• The binomial distribution and
• Recovering the average event probability from the model
• Interpreting the coefficient – the odds ratio
• Categorical predictors and the connection to the chi-square test
• Expansion to more complex models (non-linear trends, multiple predictors)
• Confounding
• Interaction terms
• Making predictions
• Comparing models and picking the ‘best’ model

Conclusion

Optional modules depending on student interest and timing:

Analyzing unstructured data with Python

• Overview of structure versus unstructured data
• Working with Unstructured Text Data
• Text data I/O with Python
• Implementing regular expressions in Python
• An overview of regular expressions
• The regex module in Python
• Regular expressions in the context of Pandas dataframes
• Converting unstructured data to structured data for analysis

Missing Data

• Exploring and understanding patterns in missing data
• Missing at Random
• Missing Not at Random
• Missing Completely at Random
• Data imputation methods