Statistics for AI & ML Developers

Learn Jupiter Notebooks and Google to run code in cells, use magic commands, and visualize data, then study descriptive statistics, measures central tendency, measures of spread, statistical tests, and hypotheses.

Learn to navigate Jupyter Notebooks in Google Colab, execute Python code in cells, and use magic commands with the percent sign while visualizing data with matplotlib.

Explore descriptive statistics to summarize data using central tendency (mean, median, mode) and spread (range, skewness, variability), clarifying population versus sample.

Examine variance and standard deviation to measure data spread, quartiles to partition data, and how null hypotheses and tests like z, t, chi-square, and p-values guide decisions.

Explore descriptive statistics, including central tendency and spread measures like mean, median, mode, range, variance, standard deviation, and quartiles, and learn about statistical tests and the null hypothesis.

Explore data distributions and inferential statistics, from probability mass and density functions to normal distributions and quantiles. Learn about sampling, p-values, and confidence intervals for hypothesis testing.

Explore how data distributions show how often values appear, compare discrete and continuous data, and use histograms, probability mass and density functions, normalization, and kernel density estimation.

Explore the normal (Gaussian) distribution, its mean and standard deviation, and how the cumulative distribution function estimates the probability of values below a threshold.

Explore quintiles and quantiles to segment data, learn the percent point function as the inverse of the cumulative distribution function, and generalize distributions beyond the normal.

Explore the student's t distribution as a more uncertain alternative to the normal distribution, compare with the uniform and exponential distributions, and understand degrees of freedom and tail behavior.

Explore the binomial distribution for two outcomes, the chi-square distribution for sums of squared standard normal variables and goodness-of-fit testing, and the F distribution for variance analysis in ANOVA.

Explore inferential statistics to draw population-level conclusions from sample data. Emphasize random unbiased sampling, p values, confidence intervals, and hypothesis testing to assess differences and relationships.

Explore probability mass and density functions, normalization, discrete versus continuous data; study distributions, cdfs, and quantiles, extend to t, uniform, exponential, chi-squared, and f distributions, with inference and confidence intervals.

Study inferential statistics, hypothesis testing, and regression to predict populations from samples, using Pearson's correlation, confidence intervals, and p-values; explore model interpretability, overfitting, bootstrapping, and Monte Carlo simulations.

Explore the line of best fit and linear regression to make predictions, using slope and intercept, and analyze correlations, covariance, and Pearson's r to balance model accuracy and interpretability.

Explore overfitting and parametric and non parametric methods to improve model generalization. Learn hypothesis testing, including significance levels, null and alternative hypotheses, and type I and II errors.

Analyze hypothesis testing with levels of significance and p values. Explore the Shibuya Shivs Inequality Theorem, the Central Limit theorem, and confidence intervals.

Learn how confidence intervals estimate a population mean using the standard normal (z) table, and how interval overlap informs hypothesis testing and the impact of sample size on precision.

Explore hypothesis testing through visualizations, bootstrapping, and Monte Carlo simulations to compare gender-based churn proportions. Build intuition with null and alternative hypotheses, p-values, and alpha to determine significance.

Explore implementing the student's t test from scratch in Python for independent and paired samples, including interpreting t statistics and p values.

Explore inferential statistics through hypothesis testing, line of best fit, and linear regression, plus correlation, overfitting, model interpretability vs accuracy, and bootstrap and Monte Carlo techniques using Python.

Compare machine learning with inferential statistics, explore the statistical basis of machine learning, and examine supervised versus unsupervised models, principal component analysis, Markov chains, transition matrices, and the bias–variance tradeoff.

Explore the differences between statistics and machine learning, focusing on inference versus prediction, and distinguish statistical models from ML models, including training, testing, and interpretability trade-offs.

Explore how statistics rests on probability spaces and loss-based estimation, while machine learning emphasizes statistical learning theory, hypothesis spaces, and model selection guided by data.

Explore practical machine learning with Python and scikit-learn, covering supervised and unsupervised learning and key algorithms. Build and evaluate a classifier on the breast cancer Wisconsin dataset using unseen data.

Explore principal component analysis as a versatile unsupervised method for dimensionality reduction, visualization, and noise filtering, using the explained variance and the first few principal components.

Explore Markov chains, their state transitions and transition matrices, and understand stationary distribution as the equilibrium, time-invariant probabilities governing long-run behavior.

Explore the bias-variance trade-off that governs generalization in supervised learning. Learn how bias and variance affect error, irreducible error, and model selection using training and test data.

Explore the statistical basis for machine learning and distinguish supervised from unsupervised models. Examine principal component analysis, Markov chains, transition matrices, and the bias-variance tradeoff.

Explore information theory concepts—entropy, joint entropy, conditional entropy, relative entropy, and cross entropy—and compare models such as random forests, k-nearest neighbors, perceptron, SVM, and Naive Bayes to predict Titanic survivors.

Explore information theory as the study of quantifying, encoding, storing, and communicating digital information, and learn how entropy measures average information, surprise, and uncertainty in random variables.

Explore joint entropy, conditional entropy, and mutual information to quantify information content between random variables, and learn about KL divergence and cross entropy as distribution differences and loss functions.

Exploratory data analysis (eda) and supervised learning approaches to predicting survival on the Titanic, with training/test splits, feature engineering, and baseline vs improved models.

Clean and prepare data by handling missing values, dropping nonessential features, engineering a deck from cabin data, and apply stochastic gradient descent to train models.

Compare random forests, logistic regression, k nearest neighbours, naive Bayes, perceptron, support vector machines, and decision trees for Titanic survival prediction, using bagging and boosting techniques.

Explore model validation with random forests, cross-validation, and hyperparameter tuning; assess performance via confusion matrices, precision, recall, the f score, ROC curves, and feature importance.

Master information theory basics—entropy, joint entropy, mutual information, conditional entropy, relative entropy, and cross entropy—and apply them to a Titanic survival project with data cleaning and stochastic gradient descent evaluation.