ChatGPT For Data Analysis: Data Science Using Python & AI

learn to import an excel file in jupyter lab using chatgpt generated python code, with pandas read_excel and file path setup for data analysis.

Develop familiarity with the Jupyter Lab notebook as a Python coding workspace. Use the file and edit menus to create, run, save, export, and manage code cells.

Master prompt engineering for ChatGPT to craft clear, effective prompts that yield precise outputs for data analysis, code development, and professional content creation.

Master ten principles of prompt engineering to craft clear, specific ChatGPT prompts for data analysis, including relating event variable to alcohol consumption via point biserial correlation.

Explore how temperature and top-k parameters shape ChatGPT responses, balancing logical precision and creative flair, with practical guidance for data analysis, coding, and API usage.

Master pivoting as a core prompt engineering principle and practice reframing prompts for complex analyses, using lifelines for Cox regression and survival data in Python.

Explore depth safety evaluation and iteration in prompt engineering, including historical context, pivoting to a library, and cross verification for data analysis.

Leverage prompt engineering to identify suitable statistical tests for two-group data, guided by ChatGPT, and learn when to use descriptive statistics, t-tests, Mann-Whitney U, regression, and ANOVA.

Discover how to use ChatGPT to frame impactful research questions and identify highly relevant topics, from cancer research to well-being of cancer patients, by prompting for future directions and scholars.

Learn how to use ChatGPT to generate a simulated dataset for factor analysis, including creating items for the big five personality scales and exporting an Excel dataset.

Explore how to use ChatGPT for manual, step-by-step calculation of factor loadings in factor analysis, including eigenvectors and eigenvalues, with caution about potential calculation errors and verification.

Generate APA style tables with ChatGPT, calculating mean, standard deviation, and 95% confidence intervals for the big five dimensions, and export in Excel or CSV.

Learn how ChatGPT can generate an APA style interpretation of a data table and how to balance AI-assisted analysis with rigorous journal writing, using effective prompts.

Build a knowledge store of statistical formulas with ChatGPT by generating a two-column quick reference table of test names and formulas, covering t-tests, ANOVA, correlation, regression, and non-parametric tests.

Master four essential data screening checks—missing values, outliers, normality, and skewness—and use ChatGPT to decide when to apply nonparametric tests instead of t tests, ANOVA, or linear regression.

Identify and analyze missing values in data sets for data analysis, compare naive approaches with imputation methods, and distinguish listwise and pairwise deletion for handling missing observations.

Explore listwise and pairwise deletion for handling missing data, illustrating data loss in listwise deletion and selective case exclusion in pairwise deletion. The lecture also introduces imputation as an alternative.

Learn how to impute missing values to retain the full data set. Explore methods like mean, median, mode, expectation maximization, regression, multiple imputation via mice, and k nearest neighbor.

Learn to perform missing value analysis with ChatGPT Plus by importing a data set, producing tabulated results, and comparing listwise and pairwise deletion with regression imputation for mean survival time.

Learn how to perform missing value analysis with ChatGPT by importing a data file, generating Python code, and replacing missing values with mean or EM algorithm, saving updated data file.

Explain how skewness reveals deviation from normality by contrasting positive skewness with a right tail and high mean to negative skewness with a left tail and high median.

Explore visual and statistical tests for skewness, including histograms, Q-q plots, and box plots, using ChatGPT to calculate skewness for income and grades (positive vs negative).

Explore Pearson's, Bowley's, and Kelly's coefficients of skewness—how to derive formulas from ChatGPT, compare mean, median and standard deviation, quartiles and percentiles, and interpret skewness in data.

Learn to compute skewness with Python via ChatGPT's regular version, importing data, selecting income and grades, and calculating Pearson's, Bowley's, and Kellie's coefficients in Jupyter Lab.

Explore normality and the normal distribution, including the bell-shaped curve and symmetry around the mean, and use the standard normal distribution with zero mean and unit variance for z-scores.

Explore the origin of the normal distribution, attributed to Gauss and Laplace, through measurement errors and the central limit theorem, and see how this Gaussian distribution shapes modern data analysis.

Explore the properties of the normal distribution: symmetry, bell-shaped curve, asymptotic tails, area under the curve equals one, and the 68.27%, 95.45%, 99.73% rules within 1–3 standard deviations.

Assess normality in a data set by applying the Kolmogorov-Smirnov and Shapiro-Wilk tests, comparing results, and noting that Shapiro-Wilk is typically more reliable.

Learn to perform normality tests on weight, age, and blood pressure with ChatGPT Plus, interpret Kolmogorov-Smirnov and Shapiro-Wilk results, visualize with histograms and Q-Q plots, and report in APA format.

Demonstrates performing normality tests in a Jupyter Lab notebook with ChatGPT, using the Kolmogorov-smirnov and Shapiro-wilk tests on weight and blood pressure, including data import and basic plots.

Explore the data analysis process, decide the setting (corporate, social, or research), and apply steps from descriptive stats to exploratory data analysis, inferential stats, and model concepts.

Explore how setting shapes data analysis across corporate, policy, and academic contexts, influencing methods, visualization, and reporting. Tailor analysis and communication to audience, balancing rigor, storytelling, and standard writing formats.

Outline steps from raw data collection to theory development, covering data cleaning, normality and outlier checks, descriptive and inferential statistics, exploratory data analysis, and model-building of constructs and concepts.

Explore how models and theories differ along scope, complexity, and testability, with examples from economics, sociology, and atomic theory, showing models are narrow and theories broad and general.

Master descriptive statistics such as mean, median, mode, and standard deviation, learn how to report them per research needs, and perform data analysis using ChatGPT.

Understand descriptive statistics as measures describing a sample and distinguish statistics from parameters, recognizing the difference between sample data and population, and the role of central tendency in summarizing attributes.

Explore how descriptive statistics locate data centers through measures of central tendency, including arithmetic mean, median, and mode, plus harmonic, geometric, and weighted means, with ChatGPT calculations.

Understand the arithmetic mean as the sample average, a central tendency measure, computed by summing observations and dividing by the sample size, with x bar notation; note population versus sample.

Compute arithmetic means for age and weight from a blood pressure data set with demographic variables like gender and age using ChatGPT Plus, upload data, and report descriptive statistics.

Analyze group differences using ChatGPT for data analysis, applying t-tests and one-way ANOVA to compare independent groups and interpret outcomes.

Explore how to choose group difference tests: independent sample t-tests for two independent groups, paired t-tests for dependent cases, and one-way or repeated measures ANOVA for multi-group comparisons.

Explore the assumptions behind parametric tests like t tests and anova, including normality, homogeneity of variance, randomness, and independence, and contrast with nonparametric alternatives.

Learn the independent sample t test: x one bar minus x two bar is numerator, pooled variance the denominator, with degrees of freedom n one plus n two minus two.

Analyze a two-group dataset with placebo and treatment, depression scores, and gender to formulate hypotheses about treatment effects and gender differences, using an independent samples t-test.

Explore calculating t-tests with ChatGPT Plus, including data preparation, assumptions, and alternatives like Mann-Whitney U, and learn to generate APA-style tables for depression score data from placebo and treatment groups.

Utilize ChatGPT and Python in Jupyter Lab to import data, run a t-test comparing placebo and treatment groups on depression scores, and generate APA style interpretation and tables.

Explore the paired sample t test and its formula using a typing speed training data set, calculating differences, standard deviation of the differences, and the t value from 50 pairs.

Formulate a paired sample t-test hypothesis using a data set of 50 participants, comparing typing speed before and after training to assess program effectiveness.

Learn to perform a paired sample t-test on pre- and post-training typing speed with ChatGPT Plus, including data import and Python code, yielding t(49)=10.95, p<0.001.

Learn to perform a paired sample t-test with ChatGPT and Python in Jupyter Lab, importing a csv file, comparing before and after typing speed, and writing an apa-style results report.

Explore one-way ANOVA with ChatGPT to compare means across three or more groups, test the null hypothesis of equal means, and apply to stress data.

Explore the theory and calculation of one-way ANOVA, including the f statistic and mean squares between and within, for comparing stress across groups.

Explore one-way ANOVA to test whether mean stress differs across three groups—college students, working professionals, and retired people—using a 120-participant dataset, formulating null and alternative hypotheses.

Conduct a one-way anova in ChatGPT to analyze stress level differences across three groups, evaluate assumptions, perform post-hoc tukey's hsd, and generate a mean plot.

Demonstrates conducting a one-way ANOVA in ChatGPT and Python to assess group differences in stress, using three groups, APA-style reporting, post-hoc tests, and mean plots.

Explore correlations, including positive and negative relationships, definitions, examples, and parametric and nonparametric methods like Pearson and Spearman, using ChatGPT to guide analysis.

Apply ChatGPT to determine the Pearson correlation coefficient between income and grades. Present a journal-ready AP style interpretation for about 1000 observations, showing a very weak, non-significant positive correlation.

Use ChatGPT 3.5 to generate code for importing data from an Excel xlsx file and computing the Pearson correlation between income and grades in a Jupyter Notebook, with APA-style interpretation.

Learn how regression analysis using ChatGPT models a dependent variable with predictors, with real estate price examples, and how to maximize predictive power through model selection and adding variables.

Explore common and less-known regression types, from linear and multiple linear regression to ridge, lasso, and polynomial methods, and apply logistic regression for categorical outcomes.

Explore polynomial regression for non-linear relationships and ridge and lasso regression for predictive modeling. Ridge handles multicollinearity among correlated predictors; lasso can shrink or zero out less important features.

Explore elastic net regression, quantile regression, and Poisson regression, showing how elastic net blends ridge and lasso for variables, how quantile regression predicts medians, and how Poisson handles count data.

Explore the assumptions of linear and multiple regression, test models, and evaluate linearity, independence of residuals, homoscedasticity, normality, and multicollinearity.

Explore linear and multiple regression using ChatGPT with a 500-person blood pressure dataset to identify whether weight, age, or gender relate to blood pressure and test the no-relationship hypothesis.

Explore regression analysis in ChatGPT using blood pressure data, with gender, weight, and age as predictors; assess linearity, homoscedasticity, normality, multicollinearity, and interpret APA style regression and ANOVA tables.

Learn to perform multiple linear regression using ChatGPT 3.5 and Python with a blood pressure data set, coding steps, and assumption checks, including APA-style reporting.