Python for Data Science and Statistical Computing

Session 7 – 02:00:00 – Numerical Computing with NumPy
This session introduces NumPy, the foundation of Python’s scientific computing stack. We begin by demonstrating why NumPy matters: performance comparisons show NumPy arrays are often 100-200x faster than Python lists for numerical operations. Participants learn to create arrays from lists, use different data types (int32, float64, etc.), and understand the advantages of homogeneous, contiguous memory layout. Array operations are covered comprehensively: element-wise operations, broadcasting rules, aggregations (sum, mean, max, min), and statistical functions. Indexing and slicing techniques include boolean indexing for filtering, fancy indexing with arrays, and reshaping operations. The session concludes with linear algebra operations including dot products and matrix multiplication. All concepts are demonstrated with scientific computing examples relevant to data analysis.

Break – 01:00:00

Session 8 – 02:00:00 – Data Processing with Pandas
This session introduces Pandas, Python’s primary library for tabular data manipulation. We begin with DataFrames and Series as the fundamental data structures, creating them from CSV files, dictionaries, and NumPy arrays. The session covers essential data manipulation operations: selecting columns and rows using various methods, filtering data with boolean indexing, using loc and iloc for explicit indexing. Data manipulation techniques include adding computed columns, sorting, and handling missing data. The powerful groupby operation is covered in detail, demonstrating the split-apply-combine paradigm for aggregations and transformations. Participants learn to compute summary statistics, perform group-wise operations, and reshape data for analysis. Practical exercises use real research datasets to develop data wrangling workflows typical of scientific data analysis.

Break – 01:00:00

Session 9 – 02:00:00 – Visualization and Statistical Analysis
This session covers data visualization and introduces statistical modeling in Python. We begin with Matplotlib for creating customizable plots with low-level control over figure elements. Seaborn is introduced for high-level statistical visualizations including distributions (histograms, KDE plots), relationships (scatter plots, regression plots), and categorical data (box plots, violin plots). The plotnine library demonstrates the grammar of graphics approach familiar to R users, allowing declarative plot construction. The session then introduces statsmodels for statistical analysis, covering basic linear regression, model fitting, and interpreting results including coefficients, p-values, and R-squared. We also cover modules and packages: importing modules, creating custom modules, and navigating Python’s package ecosystem. The session concludes with an overview of additional data science packages (scikit-learn for machine learning, scipy for scientific computing) and resources for continued learning. Participants learn to create publication-quality visualizations and perform basic statistical analyses typical of empirical research.