The qacr package contains functions and data sets designed to simplify data analyses and aid in the instruction of data science courses. The primary functions and data sets are described below.
Preparing Data
|
Function
|
Description
|
|
import
|
import() can read data from an excel spreadsheet, SAS, SPSS, or Stata data file, or a delimited text file (e.g., csv) and save it as a data frame. In the case of a delimited text file, the structure and delimiters are determined from the data.
|
|
recodes
|
recodes() provides a simple way to recode the values of numeric, character, or factor variables. See the vignette for examples.
|
|
standardize
|
standardize() transforms all the numeric variables in a data frame to same mean and standard deviation (mean=0 sd=1, by default), without modifying character, factor, or dummy coded variables.
|
|
normalize
|
normalize() transforms all the numeric variables in a data frame to same range of values ([0, 1] by default). Again, character and factor variables are left unchanged.
|
Describing a data set
|
Function
|
Description
|
|
contents
|
contents() provides a comprehensive description of a data frame. The output is much more detailed than that provided by the base summary.data.frame() function, and is easier to read and understand. This function should be your first stop when looking at a new dataset.
|
|
df_plot
|
dfPlot() helps you visual a data frame. Variable are grouped by type (numeric, integer, character, factor, date) and color coded. The percent of missing data for each variable is also displayed, along with the total number of variables and cases.
|
|
barcharts
|
barcharts() provides bar charts of all the character or factor variables in a data frame, within a single graph.
|
|
histograms
|
histograms() provides histograms of the all quantitative variables in a data frame, within a single graph.
|
|
densities
|
densities() provides density charts of all the quantitative variables in a data frame, within a single graph.
|
Exploratory data analysis
Numeric variables
|
Function
|
Description
|
|
qstats
|
qstats() allows you to easily calculate any number of descriptive statistics (e.g., n, mean, sd) for a quantiatative variable. The results can be broken down by the levels of one of more categorical variables (groups). Any function that produces a single number can be used. See the vignette for examples.
|
|
univariate_plot
|
univariatePlot() provides a detailed visualization of the distribution of values in a quantiative variable. The graph contains a histrogram, jittered dot plot, density curve, and boxplot, Annotations provide statistics such as n, mean, sd, median, min, max, skew, and outliers.
|
|
scatter
|
scatter() generates a scatter plot and line of best fit with 95% confidence interval displaying the relationship between two quantiative variables. Annotations include the slope, correlation coefficient (r), r-squared, and p_value. Oultiers (determinded by studentized residuals) are flagged. Optionally, marginal distributions (histograms, boxplots, density curves, violin plots) can be added to the margins of the plot.
|
|
cor_plot
|
corplot() plots the correlations among numeric variables in a data frame. Variables can be sorted to place variables with similar correlation patterns together.
|
|
groupdiff
|
groupdiff() compares groups on a quantitative outcome using either a parametric (ANOVA) or nonparametric (Kruskal-Wallis) test. Summary statistics, pair-wise group differences (post-hoc comparisons), and plots are provided.
|
Categorical Variables
|
Function
|
Description
|
|
tab
|
tab() generates a frequency table and bar chart for a categorical variable. There are many options including sorting categories by frequency, adding cumulative frequencies and percents, and combining infrequent categories into an ‘Other’ category. See the vignette for examples.
|
|
crosstab
|
crosstab() generates a two-way frequency table from two categorical variables. There are many options including cell, row, and column percents, plotting options, and a chi-square test of independence. See the vignette for examples.
|
Machine learning
Cluster Analysis
|
Function
|
Description
|
|
profile_plot
|
Plot and compare mean cluster profiles.
|
|
wss_plot
|
Create a within-groups sums-of-squares plot for determining the number of clusters in a numeric dataset.
|
Dimension Reduction
|
Function
|
Description
|
|
biPlot
|
Creates a principal components biplot.
|
|
FA
|
Performs common factor analysis (principle axis, maximum likelihood) with options.
|
|
PCA
|
Performs a principal components analysis, with options.
|
|
scree_plot
|
Performs a parallel analysis for determining the number of factors or compoents in a numeric dataset.
|
Classification
|
Function
|
Description
|
|
lift_plot
|
Generate gain and lift charts.
|
|
roc_plot
|
Produce a Receive Operating Curve (ROC) for a binary prediction problem. Cut-point values and AUC are also displayed.
|
Example Data sets
The qacr package contains over 25 datasets chosen to illustrate exploratory data analysis, data visualization, and machine learning. They are large enough to be interesting, while small enough to run easily on a laptop.
While each dataset can be used for data visualization, here are some pointers to get your started with specialized applications.
|
Ideas
|
Datasets
|
|
Mapping
|
Amazon forest fires, US border crossings, CIA World FactBook, Maryland crash data, US farmer’s markets, Japanese hostels, US hate crimes, California housing data, Major sports venues
|
|
Regression
|
MLB batting statistics, Boston housing data, Automobile dataset (large), Automobile dataset (small), Coffee ratings, Google play apps, Medical costs, Student grades, Time spent watching TV
|
|
Classification
|
Missed medial appointments, Breast cancer, Contraceptive use, Heart disease
|
|
Text Mining
|
British movie plots, Wine reviews
|
|
Dimension Reduction
|
Big 5 personality factors, Holland Occupational Themes
|
