README.md

In saushe/RED: Performs exploratory data analysis

RED

R Exploratory Data analysis package.

The package provides functions for comprehensive exploratory analysis.

The package automatically separates categorical and continuous variables of your data set and then performs univariate and bivariate analysis for both the types of variables. The package also contains functions to transform data to generate a model ready data set.

Description of the functions available

univariate_charvar : The function performs univariate analysis on all the categorical variables of the data. It automatically identifies all the categorical variables and performs univariate analysis

univariate_numvar : The function produces comprehensive descriptive statistics for all the numerical variables in the dataset

col_missing_count : Produces a summary of missing values for all the variables in a data set

row_missing_count : Produces a row wise summary of missing values

mv_treatment_charvar : Performs missing value treatment for character variables by imputation. Imputes missing values with 'unknown' or mode. The function also deletes the columns with missing values more than the specified cutoff

mv_treatment_numvar : Performs missing value imputation for numeric variables. Imputes missing values with mean or median. Also deletes the columns with missing values more than the specified cutoff

replace_charvars : Performs variable transformations on character variables. Groups minority classes as 'others' or replaces them with the modal class

char_bivariate : Performs bivariate analysis on the character variables. For numeric dependent variable, the function outputs mean of the dependent variable for all the classes and for categorical dependent variable the function returns crosstabs

Installation Guide

install_github() function of the devtools package can be used to install the package from github repository. First install the package devtools and then run following commands

``` R library(devtools) install_github("saushe/RED")

# Examples

Use ```help(RED)``` in ```R``` to get detailed information about the function in the package

## univariate_charvar()
```R
library(RED)
data('iris')
charvar_summary<- univariate_charvar(iris)
charvar_summary
charvar_summary$`1_Species`

Output

> charvar_summary$`1_Species`
$`#Levels`
[1] 3

$`Missing Count`
[1] 0

$Variable_Distribution
    setosa versicolor  virginica 
         1          0          0 

$`Variable_Distribution Percentage`
    setosa versicolor  virginica 
      0.01       0.00       0.00 

$`Percentage of observations in the levels with less than 5% of the observation`
[1] 0.67

univariate_numvar()

univariate_numvar(iris)

Output

           Var Count Missing Mean Median Std Dev Coeff of Variation Min Max Kurtosis Skewness 25% 75% Potential Upper Outliers Count Potential Lower Outlier Count
1 Sepal.Length             0  5.8    5.8     0.8               14.2 4.3 7.9     -0.6      0.3 5.1 6.4                              0                             0
2  Sepal.Width             0  3.1    3.0     0.4               14.3 2.0 4.4      0.1      0.3 2.8 3.3                              0                             0
3 Petal.Length             0  3.8    4.3     1.8               47.0 1.0 6.9     -1.4     -0.3 1.6 5.1                              0                             0
4  Petal.Width             0  1.2    1.3     0.8               63.6 0.1 2.5     -1.4     -0.1 0.3 1.8                              0                             0

col_missing_count()

col_missing_count(iris)

Output

[1] "0 variables have more than 50% missing values"
                      var var_type count_missing percent_missing
Sepal.Length Sepal.Length  numeric             0               0
Sepal.Width   Sepal.Width  numeric             0               0
Petal.Length Petal.Length  numeric             0               0
Petal.Width   Petal.Width  numeric             0               0
Species           Species   factor             0               0

row_missing_count ()

row_missing_count(df)

mv_treatment_numrvar()

mv_treatment_numvar(df, col.del_cutoff = 0.5) 

This will automatcally drop all the numeric columns with more than 50% missing values and wiil impute missing values in rest of the numeric columns with mean

mv_treatment_numvar(df, col.del_cutoff = 0.4, treatment_type = median)

This will drop all the numeric columns with more than 40% missing values and will impute missing values in rest of the numeric columns with median

mv_treatment_numvar(df, col.del_cutoff = 0.5, treatment_type = median, var_list = c("numvar1", "numvar2")) 

This will drop variables with more than 40% missing values but perform imputation only on variables numvar1 and numvar2

mv_treatment_charvar()

mv_treatment_charvar(df, default = T, col.del_cutoff = 0.4)

This will drop all the variables with more than 40% missing value and will impute missing value in other character variables with "unknown". Default col.del_cutoff is 0.5

 mv_treatment_charvar(df, default = T, col.del_cutoff = 0.4, char_var_list1 =  c("charvar1", "charvar2"), char_var_list2 =  c("charvar3", "charvar4")) 

Using this function, you can specify variables on which you want to perform missing value imputation. For the variables in char_var_list1, the function imputes missing values with "unknown" while for the variables in char_var_list2, the function imputes missing values with the modal class

replace_charvars()

data("cars2")
table(cars2$Country)
   France   Germany     Japan Japan/USA     Korea    Mexico    Sweden       USA 
        1         2        19         7         3         1         1        26 
 ```
 Now lets say at you want to group all the countries with less than 5% of the observations as other. You can use ```replace_charvar```

 ```R
 class(cars2$Country)
 [1] "factor"

cars2$Country<- as.character(cars2$Country) # the function does not work on factor class. I will fix it in next release

new_df<- replace_charvars(cars2, c("Country"), cutoff = 0.05) # "Classes with frequency less than the **cutoff** will be modified
table(new_df$Country)

    Japan Japan/USA     Korea     Other       USA 
       19         7         3         5        26 

If you want to replace the minority class with modal class, use the option = "rep_max"

new_df<- replace_charvars(df, c("charvar1", "charvar2)", cutoff = 0.05, option = "rep_max")

This function will will replace minority class in charvar1 and charvar2 with respective modal class

char_bivariate<- function(df, dep_var)

output1 <- char_bivariate(cars2, c("Reliability")) # Reliability is numeric variable
output2<- char_bivariate(cars2, c("Country")) # Country is categorical variable

Output

> output1$`3_Type`
  Independent_Var Dependent_Var
1         Compact      3.615385
2           Large      2.333333
3          Medium      3.090909
4           Small      4.000000
5          Sporty      2.714286
6             Van      3.666667

> output2$`2_Type`
           Type
Country     Compact Large Medium Small Sporty Van
  France          1     0      0     0      0   0
  Germany         1     0      0     1      0   0
  Japan           3     0      4     4      4   4
  Japan/USA       4     0      0     3      0   0
  Korea           0     0      1     2      0   0
  Mexico          0     0      0     1      0   0
  Sweden          1     0      0     0      0   0
  USA             5     3      8     2      5   3

Hope you find the package helpful. Please feel free to report the issues.

saushe/RED documentation built on May 29, 2019, 3:20 p.m.