| Title: | Interactive 'tourr' Using 'python' |
| Version: | 1.0.27 |
| Description: | Extends the functionality of the 'tourr' package by an interactive graphical user interface. The interactivity allows users to effortlessly refine their 'tourr' results by manual intervention, which allows for integration of expert knowledge and aids the interpretation of results. For more information on 'tourr' see Wickham et. al (2011) <doi:10.18637/jss.v040.i02> or https://github.com/ggobi/tourr. |
| License: | MIT + file LICENSE |
| Encoding: | UTF-8 |
| RoxygenNote: | 7.3.1 |
| Imports: | reticulate, tourr (≥ 1.2.4), data.table, utils |
| Suggests: | ggplot2, viridis, randomForest, knitr, rmarkdown, markdown, colorspace, ggbeeswarm, patchwork, mvtnorm, tidyverse, dplyr, gridExtra, flexclust |
| VignetteBuilder: | knitr |
| URL: | https://mmedl94.github.io/lionfish/ |
| Config/Needs/website: | knitr |
| Depends: | R (≥ 3.5.0) |
| LazyData: | true |
| NeedsCompilation: | no |
| Packaged: | 2025-03-12 17:01:22 UTC; matthias |
| Author: | Matthias Medl 
[aut, cre] |
| Maintainer: | Matthias Medl <matthias.medl@chello.at> |
| Repository: | CRAN |
| Date/Publication: | 2025-03-13 20:50:02 UTC |
Chemical Manufacturing Process Dataset
Description
Chemical Manufacturing Process Dataset
Usage
ChemicalManufacturingProcess
Format
A 176 x 58 array of continuous variables
Details
This data set contains information about a chemical manufacturing process, in which the goal is to understand the relationship between the process and the resulting final product yield and can be found in the in the AppliedPredictiveModeling R package. The data has been copied from http://appliedpredictivemodeling.com/data in agreement with their license.
Source
http://appliedpredictivemodeling.com/data
Examples
data(ChemicalManufacturingProcess)
head(ChemicalManufacturingProcess)
Australian Vacation Activities Dataset
Description
Australian Vacation Activities Dataset
Usage
ausActiv
Format
A 1003 x 44 array of binary responses
Details
The Australian Vacation Activities dataset includes responses from 1,003 adult Australians who were surveyed on their vacation activities through a permission-based internet panel in 2007.The responses are coded in binary with 0 indicating that the tourist didn't partake in the activity and 1 indicating they did.
Source
https://statistik.boku.ac.at/nachlass_leisch/MSA/
Examples
data(ausActiv)
head(ausActiv)
Check Whether 'anaconda' Environment Exists
Description
Checks if 'anaconda' environment of the given name is installed and returns TRUE if so.
Usage
check_conda_env(env_name = "r-lionfish")
Arguments
env_name |
a string that defines the name of the 'anaconda' environment reticulate uses. |
Value
boolean
Examples
check_conda_env(env_name="r-lionfish")
Check Whether Virtual 'python' Environment Exists
Description
Checks whether virtual 'python' environment of a given name exists an returns TRUE if it does.
Usage
check_venv(env_name = "r-lionfish")
Arguments
env_name |
a string that defines the name of the 'python' environment reticulate uses. |
Value
boolean
Examples
check_venv(env_name="r-lionfish")
Get Guided Tour-Holes-Better History
Description
Returns a guided tour with the holes index and the 'search_better' argument to the python' backend. This guided tour is generated with the 'tourr' functions 'save_history' and 'guided_tour'.
Usage
get_guided_holes_better_history(data, dimension)
Arguments
data |
the dataset to calculate the projections with. |
dimension |
1 for a 1d tour or 2 for a 2d tour |
Value
history object containing the projections of the requested tour
Examples
data("flea", package = "tourr")
flea <- flea[-7]
get_guided_holes_better_history(flea, 2)
Get Guided Tour-Holes History
Description
Returns a guided tour with the holes index to the python' backend. This guided tour is generated with the 'tourr' functions 'save_history' and 'guided_tour'.
Usage
get_guided_holes_history(data, dimension)
Arguments
data |
the dataset to calculate the projections with. |
dimension |
1 for a 1d tour or 2 for a 2d tour |
Value
history object containing the projections of the requested tour
Examples
data("flea", package = "tourr")
flea <- flea[-7]
get_guided_holes_history(flea, 2)
Get Guided Tour-LDA History
Description
Returns a guided tour with the LDA index to the python' backend. This guided tour is generated with the 'tourr' functions 'save_history' and 'guided_tour'.
Usage
get_guided_lda_history(data, clusters, dimension)
Arguments
data |
the dataset to calculate the projections with |
clusters |
the clusters for the lda to be performed on |
dimension |
1 for a 1d tour or 2 for a 2d tour |
Value
history object containing the projections of the requested tour
Examples
data("flea", package = "tourr")
clusters <- as.numeric(factor(flea[[7]]))
get_guided_lda_history(flea[-7], clusters, 2)
Get Local Tour History
Description
Returns a local tour based on currently displayed projection(s) to the 'python' backend. This local tour is generated with the 'tourr' functions 'save_history' and 'local_tour'.
Usage
get_local_history(data, starting_projection)
Arguments
data |
the dataset to calculate the projections with. In practice only the two first rows of the dataset are provided as the actual data is not needed. |
starting_projection |
the initial projection one wants to initiate the local tour from |
Value
history object containing the projections of the requested tour
Examples
library(tourr)
data("flea", package = "tourr")
flea <- flea[-7]
prj <- tourr::basis_random(ncol(flea), 2)
get_local_history(flea, prj)
Initialize Environment for 'python' Backend
Description
Initializes the 'python' backend required for the functionality of lionfish. At first it searches whether a 'python' environment with the provided name exists or not. If it does, it will be loaded and the 'python' function 'check_backend' is run to check if it works properly. If no 'python' environment with the provided name exists, it will be installed and then loaded. This can either be done with or without Anaconda as package manager. Anaconda can be more robust, but the GUI will appear dated. Thus, trying 'init_env' with virtual_env = "virtual_env" out first is recommended. For 'Windows' users, the path to the tk and tcl libraries will be set, otherwise tkinter cannot run.
Usage
init_env(env_name = "r-lionfish", virtual_env = "virtual_env", local = FALSE)
Arguments
env_name |
a string that defines the name of the python environment reticulate uses. This can be useful if one wants to use a preinstalled python environment. |
virtual_env |
either "virtual_env" or "anaconda". "virtual_env" creates a virtual environment, which has the advantage that the GUI looks much nicer and no previous python installation is required,but the setup of the environment can be more error prone. "anaconda" installs the python environment via Anaconda, which can be more stable, but the GUI looks more dated. |
local |
logical |
Value
initializes python environment
Examples
if (check_venv()){
init_env(env_name = "r-lionfish", virtual_env = "virtual_env")
} else if (check_conda_env()){
init_env(env_name = "r-lionfish", virtual_env = "anaconda")
}
R Wrapper for 'interactive_tour' Function Written in 'python'
Description
Launches the lionfish GUI and at minimum requires the data do be loaded and the plot_objects. The other parameters are optional. For technical reasons the parameters half_range, n_plot_cols, n_subsets, color_scale, label_size and display_size cannot be adjusted from within the GUI. The GUI has to be closed and relaunched (possibly with load_interactive_tour) if you want to change them. Please visit https://mmedl94.github.io/lionfish/index.html for a detailed description of the GUI and its features.
Usage
interactive_tour(
data,
plot_objects,
feature_names = NULL,
half_range = NULL,
n_plot_cols = 2,
preselection = FALSE,
preselection_names = FALSE,
n_subsets = 3,
display_size = 5,
hover_cutoff = 10,
label_size = 15,
color_scale = "default",
color_scale_heatmap = "default",
axes_blendout_threshhold = 1
)
Arguments
data |
the dataset you want to investigate |
plot_objects |
a named list of objects you want to be displayed. Each entry requires a definition of the type of display and a specification of what should be plotted. |
feature_names |
names of the features of the dataset |
half_range |
factor that influences the scaling of the displayed tour plots. Small values lead to more spread out datapoints (that might not fit the plotting area), while large values lead to the data being more compact. If not provided a good estimate will be calculated and used. |
n_plot_cols |
specifies the number of columns of the grid of the final display. |
preselection |
a vector that specifies in which subset each datapoint should be put initially. |
preselection_names |
a vector that specifies the names of the preselection subsets |
n_subsets |
the total number of available subsets. |
display_size |
rough size of each subplot in inches |
hover_cutoff |
number of features at which the switch from intransparent to transparent labels that can be hovered over to make them intransparent occurs |
label_size |
size of the labels of the feature names of 1d and 2d tours |
color_scale |
a viridis/matplotlib colormap to define the color scheme of the subgroups |
color_scale_heatmap |
a viridis/matplotlib colormap to define the color scheme of the heatmap |
axes_blendout_threshhold |
initial value of the threshold for blending out projection axes with a smaller length |
Value
opens the interactive GUI
Examples
library(tourr)
data("flea", package = "tourr")
data <- flea[1:6]
clusters <- as.numeric(flea$species)
flea_subspecies <- unique(flea$species)
feature_names <- colnames(data)
guided_tour_history <- tourr::save_history(data,
tour_path = tourr::guided_tour(holes())
)
grand_tour_history_1d <- tourr::save_history(data,
tour_path = tourr::grand_tour(d = 1)
)
half_range <- max(sqrt(rowSums(data^2)))
obj1 <- list(type = "2d_tour", obj = guided_tour_history)
obj2 <- list(type = "1d_tour", obj = grand_tour_history_1d)
obj3 <- list(type = "scatter", obj = c("tars1", "tars2"))
obj4 <- list(type = "hist", obj = "head")
if (check_venv()){
init_env(env_name = "r-lionfish", virtual_env = "virtual_env")
} else if (check_conda_env()){
init_env(env_name = "r-lionfish", virtual_env = "anaconda")
}
if (interactive()){
interactive_tour(
data = data,
plot_objects = list(obj1, obj2, obj3, obj4),
feature_names = feature_names,
half_range = half_range,
n_plot_cols = 2,
preselection = clusters,
preselection_names = flea_subspecies,
n_subsets = 5,
display_size = 5
)
}
R Wrapper for 'load_interactive_tour' Function Written in 'python'
Description
Loads a previously saved snapshot created by pressing the "Save projections and subsets" within the GUI. The data that was loaded when saving a snapshot has to be provided to this function when loading that snapshot. Additionally, this function allows to adjust some parameters of the GUI when loading a snapshot, such as display_size or label_size.
Usage
load_interactive_tour(
data,
directory_to_save,
half_range = NULL,
n_plot_cols = 2,
preselection = FALSE,
preselection_names = FALSE,
n_subsets = FALSE,
display_size = 5,
hover_cutoff = 10,
label_size = 15,
color_scale = "default",
color_scale_heatmap = "default",
axes_blendout_threshhold = 1
)
Arguments
data |
the dataset you want to investigate. Must be the same as the dataset that was loaded when the save was created! |
directory_to_save |
path to the location of the save folder |
half_range |
factor that influences the scaling of the displayed tour plots. Small values lead to more spread out datapoints (that might not fit the plotting area), while large values lead to the data being more compact. If not provided a good estimate will be calculated and used. |
n_plot_cols |
specifies the number of columns of the grid of the final display. |
preselection |
a vector that specifies in which subset each datapoint should be put initially. |
preselection_names |
a vector that specifies the names of the preselection subsets |
n_subsets |
the total number of available subsets (up to 10). |
display_size |
rough size of each subplot in inches |
hover_cutoff |
number of features at which the switch from intransparent to transparent labels that can be hovered over to make them intransparent occurs |
label_size |
size of the labels of the feature names of 1d and 2d tours |
color_scale |
a viridis/matplotlib colormap to define the color scheme of the subgroups |
color_scale_heatmap |
a viridis/matplotlib colormap to define the color scheme of the heatmap |
axes_blendout_threshhold |
initial value of the threshold for blending out projection axes with a smaller length |
Value
opens the interactive GUI
Examples
data("flea", package = "tourr")
data <- flea[1:6]
if (check_venv()){
init_env(env_name = "r-lionfish", virtual_env = "virtual_env")
} else if (check_conda_env()){
init_env(env_name = "r-lionfish", virtual_env = "anaconda")
}
pytourr_dir <- find.package("lionfish", lib.loc = NULL, quiet = TRUE)
pytourr_dir <- paste(pytourr_dir, "/inst/test_snapshot", sep = "")
if (interactive()){
load_interactive_tour(data, pytourr_dir)
}
Modification of the 'render_proj' Function of 'tourr'
Description
Modification of the render_proj() function of tourr so that the half_range is calculated with max(sqrt(rowSums(data^2))) or can be provided as argument.
Usage
render_proj_inter(
data,
prj,
half_range = NULL,
axis_labels = NULL,
obs_labels = NULL,
limits = 1,
position = "center"
)
Arguments
data |
matrix, or data frame containing numeric columns, should be standardized to have mean 0, sd 1 |
prj |
projection matrix |
half_range |
for scaling in the display, by default calculated from the data |
axis_labels |
of the axes to be displayed |
obs_labels |
labels of the observations to be available for interactive mouseover |
limits |
value setting the lower and upper limits of projected data, default 1 |
position |
position of the axes: center (default), bottomleft or off |
Value
list containing projected data, circle and segments for axes
Examples
library(tourr)
data("flea", package = "tourr")
flea_std <- apply(flea[,1:6], 2, function(x) (x-mean(x))/sd(x))
prj <- tourr::basis_random(ncol(flea[,1:6]), 2)
p <- render_proj_inter(flea_std, prj)
Risk Dataset
Description
Risk Dataset
Usage
risk
Format
A 563 x 6 array of Likert scale data
Details
Adult Australian residents that have undertaken at least one holiday in the last year, which involved staying away from home for at least four nights, were asked what risks they have taken in the past. The questions were about recreational, health, career, financial, safety and social risk. The response options were on the following: never (1), rarely (2), quite often (3), often (4) or very often (5)
Source
https://statistik.boku.ac.at/nachlass_leisch/MSA/
Examples
data(risk)
head(risk)
Austrian Vacation Activities Dataset
Description
Austrian Vacation Activities Dataset
Usage
winterActiv
Format
A 2961 x 27 array of binary responses
Details
The Austrian Vacation Activities dataset comprises responses from 2,961 adult tourists who spent their holiday in Austria during the 1997/98 season. The responses are coded in binary with 0 indicating that the tourist didn't partake in the activity and 1 indicating they did.
Source
https://statistik.boku.ac.at/nachlass_leisch/MSA/
Examples
data(winterActiv)
head(winterActiv)