Statistics & Mathematics for Data Science in Python

Explore descriptive statistics and the exploratory stages of bringing in data, build statistical vocabulary, classify variable types, and compute counts, measures of center, standard deviation, variance, and correlation coefficients.

Learn general statistical vocabulary for data science in Python, including population versus sample, categorical and quantitative data, bias, interviewer effects, lurking variables, data collection, cleaning, and basic analysis.

Learn to load web data with pandas, read HTML tables, select a specific table, inspect variable types within data, clean data types, and create new data frames for analysis.

Load data from URL pages with pandas read_html, then summarize categorical data using counts to reveal trends. Concatenate data frames to build unified datasets and apply value_counts for quick exploration.

Master measures of center and essential statistics in Python using pandas. Import data from a URL, convert types to numeric, and compute mean, median, std, var, and describe.

Explore the correlation coefficient, distinguish correlation from causation, and interpret strength in the -1 to 1 range, with practical examples and notes on calculation.

Apply statistical vocabulary to explore data and present findings using pandas to read tables from URLs, perform scraping, calculate measures of center, deviation, and correlation, and understand correlation versus causation.

Explore probability distributions, learn distribution characteristics, and generate sample distributions in Python using Google Colab, then visualize histograms to identify distribution types from their shapes and key characteristics, with code.

Explore probability distributions by distinguishing discrete and continuous random variables, identifying their possible outcomes, and understanding how these shapes and characteristics inform predictive analytics and margin of error.

Identify key features of the uniform distribution, where every outcome is equally likely, explore shape, generate data that fits it, and visualize it with histograms and kernel density estimate plots.

Explore the binomial distribution, two possible outcomes, and subjective success. See how the number of trials and probability shape the distribution, with visualization and the expected value.

Identify the Poisson distribution's key features and lambda-driven shapes, generate Poisson data in Python, and explore probabilities of events within a fixed time interval through visualizations.

Explore the normal distribution, a bell-shaped model for continuous data, and learn to use mean and standard deviation to describe, simulate, and visualize data in exploratory data analysis.

Explore continuous distributions with real data using fitter to compare fits and identify the best distribution by least-squares error. Understand how non normal data and outliers affect distribution choice.

Apply distribution characteristics, generate sample distributions in Python from all covered distributions, and create visualizations for exploratory analysis of uniform, binomial, Poisson, and normal distributions.

Explore inferential statistics with visualizations; learn to choose appropriate visuals for categorical and quantitative data, and interpret bar charts, histograms, box plots, and scatter plots for exploratory data analysis.

Explore how bar charts reveal data type suitability, show group differences, and visualize categorical variables using seaborn catplot with Titanic and iris data, including hue, column, and row splits.

Explore how histograms reveal the distribution of quantitative data and enable comparing distributions using bins, rug marks, color, and labels, while noting their limitations.

Explore box plots for quantitative data, identify outliers, and compare distributions with Titanic age and iris species examples to reveal interpretation and limitations.

Explore scatterplots to compare two quantitative variables, interpret patterns and correlations, and use regression lines with R-squared and confidence intervals, plus customization options.

Explore advanced visualizations in Python data science, combining categorical and quantitative data with lm plots, joint plots, hue, markers, and palettes; compare distributions with kde plots and annotate Pearson correlations.

Determine appropriate visualizations for categorical and quantitative data, and create bar charts, histograms, box plots, scatter plots, and advanced lm plots and joint plots to explore multivariate data and correlations.

Explore confidence intervals and hypothesis testing, interpret statistical significance, and apply normal approximation techniques and visualization techniques; compare groups with anova, chi-squared, and bootstrap for non-normal distributions.

Explore Seaborn sample data types and data frames for visualization and machine learning in Python. Use joint plots and the fitter to compare distributions, assess normality, and handle non-normal data.

Explore how to compute and interpret confidence intervals and hypothesis tests in data science with Python, including means vs proportions, margin of error, and p-value for statistical significance.

Explore why assuming normality matters for statistical testing, compare standardizing versus normalizing data, and apply t scores to standardize distributions like bill and tip data.

Explore normal data analysis by creating and interpreting a normal probability plot, assessing normality with R-squared, and comparing standardized versus nonstandardized data using tips and total bill datasets.

Explore normal data and categorical confidence intervals, comparing two samples to test differences in proportions. Use 95 percent confidence intervals and z tests to assess p-values and statistical significance.

Master normal data and quantitative confidence intervals, including z and t intervals. Learn Welch’s, paired, and two-sample t-tests and hypothesis testing with the diamonds data set.

ANOVA compares means across more than two groups by analyzing variances and uses the statsmodels and the Pinguin library to assess normality, equality of variances, independence, and perform pairwise comparisons.

Demonstrate non normal data handling with bootstrap to estimate means and confidence intervals, compare to the t distribution, and discuss sampling assumptions, replacement, and block bootstrapping.

Master statistical significance, interpret a confidence interval, and conduct a hypothesis test while analyzing group differences in proportions and means under normality assumptions, using anova, chi-square, and bootstrap techniques.

Learn how regression and predictions drive data science, including cleaning and preparing data, building predictive models, and hyper tuning to compare models and optimize predictions from multiple quantitative variables.

Explore how to load and explore the diamonds data, visualize distributions with Seaborn, and prepare data for modeling, including train-test splits and regression and classification approaches.

Clean and prepare data by identifying logical data issues, removing weird zeros, and justifying edits; encode categoricals with one-hot encoding and perform a train-test split for model evaluation.

Explore linear regression as a simple, efficient baseline in Python, fitting multi-variable models with one-hot encoding and evaluating with absolute error, mean squared error, and R-squared.

Explore polynomial regression, its pros and cons, implement it with polynomial features of degrees 3 and 5 on a data set, and compare performance via r-squared and model summary.

Explore ridge regression, its regularization strength alpha, and built-in cross-validation to balance bias and variance when predictors exceed observations; compare to linear regression and interpret outputs.

Explore lasso regression, a shrinkage-based linear model that selects features, shrinks coefficients toward zero, and avoids overfitting. Compare its bias and performance with ridge using cross-validated alphas and error metrics.

Explore elastic net regression, compare its bias-variance tradeoffs with ridge and lasso, interpret outputs, and assess predictive performance for diamond prices using cross-validated alpha and L1 ratios.

Learn random forest regression, its bias reduction with multiple trees, and robustness to missing data and categorical features. See a Google Colab notebook implementation and how estimator counts affect R-squared.

Compare multiple regression models using a cross-validated r-squared based tool, visualize results with a box and strip plot, and identify random forest regressor as top performer.

Apply grid search to tune a random forest regressor and optimize model performance using cross-validation, examining estimators, max depth, and min samples for split to maximize r-squared.

Clean and prepare data for analysis, apply prediction and regression techniques, and build and compare predictive models using cross-validation visuals and hyperparameter tuning to reveal insights.

Explore classification modeling with quantitative and categorical data, loading and cleaning data, training naive Bayes, logistic regression, k-nearest neighbors, SVM, and random forest models, and compare performance with hyperparameter tuning.

Load and explore penguins data, inspect types and distributions, remove missing values to prepare data frames for classification models, and visualize patterns with seabourne package and box plots to guide training.

Remove null values, drop the island column, balance species samples, and split the data into training and testing sets for classification analysis.

Explore Naive Bayes, including its probabilistic basis and a multinomial Naive Bayes classifier, and interpret accuracy using a confusion matrix heat map and a classification report.

Learn how logistic regression classifies data by predicting probabilities between zero and one, with multinomial extensions for multiple groups, and evaluate performance using confusion matrices and classification reports.

Learn how the K nearest neighbors algorithm classifies data by clustering around similar points, compare its accuracy to logistic and naive Bayes, and assess results with a confusion matrix.

Explore support vector machines and how a hyperplane separates classes, using outliers near the boundary to improve separation; compare with logistic regression and k-nearest neighbors.

Learn how the random forest classifier classifies data, trains an RF model, and compares its accuracy against logistic, SVM, and Naive Bayes using a confusion matrix.

Compare logistic regression, k-nearest neighbors, support vector classifier (svc), and random forest using a model comparison tool with visualizations of accuracy across cross-validation.

Apply model hyper tuning and optimization to classification and regression, using grid search on multinomial naive Bayes with alpha and fit prior, five-fold cross-validation, and best-accuracy reporting.

Load, clean, and finalize analysis data for classification tasks. Implement Naive Bayes, logistic regression, k-nearest neighbors, and random forest models, compare performance with cross-validation, and visualize results.

Explore natural language processing techniques in Python, including text processing in Colab notebooks, visualizing text frequency, and applying naïve Bayes classification and cosine similarity for sentiment and context analysis.

Load and explore data in Python for data science, fetch HTML text with requests and BeautifulSoup, clean and visualize it, then prepare it in pantos data frames for analysis.

Learn to process text data with nltk by tokenizing, lowering case, removing stop words, and visualizing word frequencies to reveal the main topics in a dataset.

Create a single large for loop to load data from Gutenberg texts, process, visualize, and display it, building a frequency distribution with tokenization and stop-word filtering.

Explore movie reviews text analysis using a movie reviews corpus, sort data by category, compute word frequencies, identify most common words, and demonstrate stop words removal.

Identify the top 2500 word features in positive and negative documents, mark their presence by sentiment, and build a features set to analyze textual data.

Apply naive bayes with nltk to build a sentiment classifier from a 75/25 train-test split, measure accuracy, and reveal the most informative features separating positive and negative reviews.

learn to compute cosine similarity between texts with tf-idf vectorization, clean and tokenize four Gutenberg texts, and build a length-insensitive similarity matrix.

Load text data into a Google Colab notebook, analyze textual features, visualize text frequency, and apply naive bayes with cosine similarity across four texts.