1 Why you should care about statistics

Statistics is the discipline of describing and inferring truths from data. It treats the data you have as a sample from a larger population and helps you measure uncertainty, test ideas, and make better decisions. Because data is everywhere, statistical literacy is a timeless, cross-industry advantage.

• Employability: Combine domain knowledge with inference to uncover valuable signals.
• Data utility: Turn raw data into actionable insights instead of leaving it unused.
• Decision-making: Quantify uncertainty to guide choices under risk.
• Machine learning/AI: Build and evaluate models with a stronger statistical mindset.
• Effective sampling: Design better experiments and interpret samples vs. populations.

Instead of memorizing lookup tables and mechanics first, the book emphasizes intuition and real-world examples, then uses simple Python functions to do the heavy lifting. You focus on the big picture, not arcane calculations.

• Relate samples to populations and apply this to modern applications like ML.
• Compute descriptive stats: mean, median, mode, variance, standard deviation, interquartile range, proportions.
• Infer with confidence intervals and hypothesis tests for means, proportions, and variances.
• Run linear and logistic regression and understand how statistical and ML approaches differ.

• Basics: syntax, variables, functions/parameters, if/elif, for loops, importing packages.
• Helpful libraries: numpy, pandas, matplotlib.
• Environment: any Python 3 setup works (e.g., VS Code, PyCharm, Jupyter/Colab, Anaconda). The code footprint is small and tooling is kept minimal.

• Hypothesize: Form a question or claim to investigate.
• Gather data: Obtain a sample relevant to the hypothesis.
• Fit a model: Choose an appropriate model (e.g., linear regression) to quantify relationships.
• Test/evaluate: Check performance on new data and measure errors/uncertainty.

Example: Hypothesize that temperature affects sports drink sales, fit a linear model, then test on unseen data to validate or refine the idea.

Statistics turns noisy history into actionable guidance. For example, when planning inventory you can:

• Use time series to detect seasonality (e.g., holiday spikes).
• Run hypothesis tests to assess whether discounts or ads worked.
• Apply regression to link ad spend or design changes to conversion.

The goal is not perfect prediction, but measured uncertainty that supports better decisions.

• Check sampling: Who was measured, and is it representative?
• Inspect methods: Assumptions, p-values, multiple comparisons, outlier handling.
• Look for confounders: Alternative explanations for the observed effect.
• Follow incentives: Who funded the work, and what outcomes are rewarded?

A skeptical, statistics-literate eye helps you reverse-engineer claims and get closer to the truth.

Misaligned incentives can drive cherry-picking, “data torturing,” and burying unfavorable results. Academic “publish or perish,” sponsored research, and organizational pressures can bias analyses. Knowing proper methods—and being diplomatic—helps you navigate and advocate for integrity.

• Overlap: Both learn from data to explain or predict.
• Emphasis: Statistics prioritizes explainability and uncertainty; ML emphasizes predictive performance and optimization, often via black-box models.
• Pragmatics: For high-dimensional problems (images, language), black-box ML may be necessary; still, use statistics to validate outputs and assess bias.
• Rule of thumb: Prefer statistical approaches when interpretability and inference matter; use ML when large-scale prediction is the primary goal.