Data Analysis Workflows and R Programming Guide
Data Workflow and the LOCKMES Framework
- Workflows consist of discrete, sequential steps that help organizations accomplish tasks and generate value.
- Workflows are visually represented outside of class through flowcharts, diagrams, and checklists.
- The structured approach to data analysis in R follows the acronym LOCKMES:
- L — Look at your data:
- Inspect the structure, dimensions, number of variables, and number of observations before performing analytical or statistical routines.
- Identify the presence of missing data prior to executing code.
- C — Clean the data:
- Evaluate text and font properties within the dataset.
- Rename column titles to appropriate formats.
- Check for and eliminate unexpected or corrupted characters introduced during data import.
- K / M — Handle Missing data:
- Identify mechanisms and classifications of missing values rather than automatically deleting incomplete rows or cells.
- Complete formal data imputation or processing workflows.
- E — Explore the data through visualizations:
- Construct graphical representations including histograms, box plots, bar charts, and frequency tables to observe data distributions.
- S — Statistics:
- Calculate summary measures such as means, counts, and proportions.
- S — Summarize findings:
- Draft clear, well-written analytical statements that contextualize statistical results, recognizing that writing is a core component of statistics.
Best Practices in R Programming
- Object Assignment:
- Avoid hard-coding constant numeric values directly into long script computations.
- Assign computed values, vectors, and imported files to objects using assignment notation (
<-) to increase computational efficiency and script reproducibility.
- Regular Data Inspection:
- Inspect output continuously throughout coding to confirm functions execute as intended and variables are not modified inadvertently.
- Dataset Preservation:
- Never make direct alterations to an original dataset.
- Create working copies of datasets prior to modifying structure or creating derived variables to preserve raw baseline records.
- Code Documentation:
- Use comments (
#) extensively to annotate code, detail analytical decisions, and document functional logic for external collaborators and future reference.
Data Types and Structural Inspection
- Univariate Data Classifications:
- Categorical Data: Visualized using pie charts, bar graphs, frequency tables, and relative frequency tables.
- Quantitative Data: Visualized using histograms and box plots.
- Metadata:
- Variable descriptions and metadata provide explanatory background information detailing context, variable definitions, and measurement units.
- Data Dimensions:
- Observations: Represented by dataset rows (n).
- Variables: Represented by dataset columns.
- Data Inspection Functions in R:
str(data_frame): Primary command for initial dataset assessment. Displays total observation count (rows), variable count (columns), column names, data classifications (e.g., numeric, character), and initial values.dim(data_frame): Displays precise dimensions of a data frame returning row count followed by column count.
- Missing Value Detection:
- Missing values in R are denoted by the system logical indicator
NA. is.na(data_frame): Evaluates every cell in a dataset, returning a logical output (TRUE for missing, FALSE for non-missing).sum(is.na(data_frame)): Sums the logical outputs where TRUE equals 1 and FALSE equals 0, returning the exact total count of NA values across the dataset.
Data Visualizations in R
Histograms
- Base Syntax:
hist(data_frame$variable) generates a default histogram with automated binning and a gray color scheme. - Essential Graphical Components:
- Every graph must include an explicit title and labeled axes with units.
- Histogram Arguments: