Motivation

I want to make building the best plots as easy as possible. I've never really been a fan of incrementally building a plot by calling function after function, mostly because I have to keep going to stackoverflow to get the syntax or flip through entire documentation just to see what's possible.

This package is intended to reduce or eliminate that behavior (hence the "Auto" part of the name "AutoPlots"). The plots returned in AutoPlots are sufficiently good for 99% of plotting purposes. There are two broad classes of plots available in AutoPlots: Standard Plots and Model Evaluation Plots. If other users find additional plots that this package can support I'm open to having them incorporated.

Standard plots

Histogram Plots
Density Plots
Box Plots
Probability Plots
Word Cloud
Pie Charts
Donut Plot
Rosetype Plot
Bar Plots
3D Bar Plots
Stacked Bar Plots
Radar Plots
Line Plots
Step Plots
Area Plots
River Plots
Autocorrelation Plot
Partial Autocorrelation Plot
Scatter Plots
3D Scatter Plots
Copula Plots
3D Copula Plots
Correlation Matrix Plots
Parallel Plots
Heatmaps

Model evaluation

These plot types are most useful for those looking to evaluate the performance of regression, binary classification, and multiclass models. Designing plots for multiclass models are rather challenging but I've abstracted all that work away so the user only has to pass their categorical target variable along with their categorical predicted value, and the plots will display all the levels appropriately without requiring the user to do the data manipulation ahead of time. Same goes for regression and classification, which are easier, but still requires time and energy.

Additionally, all model evaluation plots supports grouping variables for by-analysis of models, even for multiclass models!

Calibration Plots
Calibration Scatter Plots
Partial Dependence Plots
Partial Dependence Heatmaps
Variable Importance Plots
Shapely Importance Plots
ROC Plots
Confusion Matrix Heatmaps
Lift Plots
Gain Plots

Data Management

Another giant bonus is that the user can either pre-aggregate their data and pass that through to these functions (using PreAgg = TRUE) or they can leave their data in raw form and let my optimized data.table code manage it for them. This means you can develop plots from giant data sets without having to wait for long running data operations. Further, there is a SampleSize parameter in the functions to limit the number of records to display, for the giant data cases (or for scatter / copula plots). This sampling takes place AFTER data aggregation, not before.

Features

Common API across all functions, regardless of Echarts usage or Plotly usage
Automatic data management via data.table operations
Large variety of aggregation statistics options
Large number of numeric transformations options
Easy faceting by specifying FacetRows and FacetCols via function parameters
Automatic formatting from Echarts
Updating Titles, Axes Labels, and Values displayed on plots
There are 37 plot types (25 standard and 12 model evaluation) including 3D Plots
Display size sampling (sampled right before plot building, not before data management)
Model evaluation plots available by grouping variables (or faceted)

Getting Started

Installation from CRAN

install.packages("AutoPlots")

Installation from GitHub

install.packages("combinat")
install.packages("data.table")
install.packages("devtools")
install.packages("dplyr")
install.packages("e1071")
install.packages("echarts4r")
install.packages("lubridate")
install.packages("nortest")
install.packages("quanteda")
install.packages("quanteda.textstats")
install.packages("scales")
install.packages("stats")
install.packages("utils")
devtools::install_github("AdrianAntico/AutoPlots", upgrade = FALSE, force = TRUE)

Sample of Plot Images

Histogram

# Create fake data
data <- AutoPlots::FakeDataGenerator(N = 1000000)

# Build Histogram plot
AutoPlots::Plot.Histogram(
  dt = data,
  XVar = NULL,
  YVar = "Independent_Variable4",
  YVarTrans = "Identity",
  EchartsTheme = "macarons")

Density

# Create fake data
data <- AutoPlots::FakeDataGenerator(N = 1000000)

# Build Density plot
AutoPlots::Plot.Density(
  dt = data,
  XVar = NULL,
  YVar = "Independent_Variable4",
  YVarTrans = "Identity",
  EchartsTheme = "macarons")

Box Plot

# Create fake data
data <- AutoPlots::FakeDataGenerator(N = 1000000)

# Build Box plot
AutoPlots::Plot.Box(
  dt = data,
  XVar = "Factor_1",
  YVar = "Independent_Variable1",
  YVarTrans = "Identity",
  EchartsTheme = "macarons")

Pie Plot

# Create fake data
data <- AutoPlots::FakeDataGenerator(N = 1000000)

# Build Pie plot
AutoPlots::Plot.Pie(
  dt = data,
  XVar = "Factor_1",
  YVar = "Independent_Variable1",
  YVarTrans = "Identity",
  EchartsTheme = "macarons")

Area Plot

# Create fake data
data <- AutoPlots::FakeDataGenerator(N = 1000)

# Build Area plot
AutoPlots::Plot.Area(
  dt = data,
  PreAgg = FALSE,
  AggMethod = "mean",
  XVar = "DateTime",
  YVar = "Independent_Variable1",
  YVarTrans = "Identity",
  EchartsTheme = "macarons")

Line Plot

# Create fake data
data <- AutoPlots::FakeDataGenerator(N = 1000)

# Build Line plot
AutoPlots::Plot.Line(
  dt = data,
  PreAgg = FALSE,
  AggMethod = "mean",
  XVar = "DateTime",
  YVar = "Independent_Variable1",
  YVarTrans = "Identity",
  EchartsTheme = "macarons")

Step Plot

# Create fake data
data <- AutoPlots::FakeDataGenerator(N = 1000)

# Build Step plot
AutoPlots::Plot.Step(
  dt = data,
  PreAgg = FALSE,
  AggMethod = "mean",
  XVar = "DateTime",
  YVar = "Independent_Variable1",
  YVarTrans = "Identity",
  EchartsTheme = "macarons")

River Plot

# Create fake data
data <- AutoPlots::FakeDataGenerator(N = 1000)

# Build River plot
AutoPlots::Plot.River(
  dt = data,
  PreAgg = FALSE,
  AggMethod = "mean",
  XVar = "DateTime",
  YVar = c(
    "Independent_Variable1",
    "Independent_Variable2",
    "Independent_Variable3",
    "Independent_Variable4",
    "Independent_Variable5"),
  YVarTrans = "Identity",
  EchartsTheme = "macarons")

Bar Plots

# Create fake data
data <- AutoPlots::FakeDataGenerator(N = 100000)

# Echarts Bar Chart
AutoPlots::Plot.Bar(
  dt = data,
  PreAgg = FALSE,
  XVar = "Factor_1",
  YVar = "Adrian",
  YVarTrans = "Identity",
  EchartsTheme = "macarons")

Stacked Bar Plots

# Create fake data
data <- AutoPlots::FakeDataGenerator(N = 100000)

# Echarts Stacked Bar Chart
AutoPlots::Plot.StackedBar(
  dt = data,
  PreAgg = FALSE,
  XVar = "Factor_1",
  YVar = "Adrian",
  GroupVar = "Factor_2",
  YVarTrans = "Identity",
  EchartsTheme = "macarons")

3D Bar Plots

# Create fake data
data <- AutoPlots::FakeDataGenerator(N = 100000)

# Echarts 3D Bar Chart
AutoPlots::Plot.BarPlot3D(
  dt = data,
  PreAgg = FALSE,
  XVar = "Factor_1",
  YVar = "Factor_2",
  ZVar = "Adrian",
  YVarTrans = "Identity",
  EchartsTheme = "macarons")

Scatter Plot

# Create fake data
data <- AutoPlots::FakeDataGenerator(N = 100000)

# Echarts Scatter Plot Chart
AutoPlots::Plot.Scatter(
  dt = data,
  SampleSize = 10000,
  XVar = "Adrian",
  YVar = "Independent_Variable8",
  YVarTrans = "Identity",
  EchartsTheme = "macarons")

3D Scatter Plot

# Create fake data
data <- AutoPlots::FakeDataGenerator(N = 100000)

# Echarts Scatter Plot Chart
AutoPlots::Plot.Scatter3D(
  dt = data,
  SampleSize = 10000,
  XVar = "Adrian",
  YVar = "Independent_Variable8",
  ZVar = "Independent_Variable6",
  YVarTrans = "Identity",
  EchartsTheme = "macarons")

Copula Plot

# Create fake data
data <- AutoPlots::FakeDataGenerator(N = 100000)

# Echarts Copula Plot Chart
AutoPlots::Plot.Copula(
  dt = data,
  SampleSize = 10000,
  XVar = "Adrian",
  YVar = "Independent_Variable8",
  YVarTrans = "Identity",
  EchartsTheme = "macarons")

3D Copula Plot

# Create fake data
data <- AutoPlots::FakeDataGenerator(N = 100000)

# Echarts Copula Plot Chart
AutoPlots::Plot.Copula3D(
  dt = data,
  SampleSize = 10000,
  XVar = "Adrian",
  YVar = "Independent_Variable9",
  ZVar = "Independent_Variable6",
  YVarTrans = "Identity",
  EchartsTheme = "macarons")

Heatmap Plot

# Create fake data
data <- AutoPlots::FakeDataGenerator(N = 100000)

# Echarts Heatmap Plot Chart
AutoPlots::Plot.HeatMap(
  dt = data,
  XVar = "Factor_1",
  YVar = "Factor_2",
  ZVar = "Independent_Variable6",
  YVarTrans = "Identity",
  EchartsTheme = "macarons")

CorrMatrix Plot

# Create fake data
data <- AutoPlots::FakeDataGenerator(N = 100000)

# Echarts CorrMatrix Plot Chart
AutoPlots::Plot.CorrMatrix(
  dt = data,
  CorrVars = c(
    "Adrian",
    "Independent_Variable1",
    "Independent_Variable2",
    "Independent_Variable3",
    "Independent_Variable4",
    "Independent_Variable5"),
  EchartsTheme = "macarons")

Probability Plot

dt <- data.table::data.table(Y = pnorm(q = runif(10000)))
AutoPlots::Plot.ProbabilityPlot(
  dt = dt,
  SampleSize = 1000L,
  YVar = "Y",
  YVarTrans = "Identity",
  Height = NULL,
  Width = NULL,
  Title = 'Normal Probability Plot',
  ShowLabels = FALSE,
  EchartsTheme = "blue",
  TextColor = "black",
  title.fontSize = 22,
  title.fontWeight = "bold",
  title.textShadowColor = '#63aeff',
  title.textShadowBlur = 3,
  title.textShadowOffsetY = 1,
  title.textShadowOffsetX = -1,
  yaxis.fontSize = 14,
  yaxis.rotate = 0,
  ContainLabel = TRUE,
  tooltip.trigger = "axis",
  Debug = FALSE)

indykpol / autoplots Goto Github PK

autoplots's Introduction