Identify Penguin species

Identify penguin species using multinomial Logistic Regression

analysis
R
Author

Aditya Ranade

Published

April 29, 2025

I found this dataset on UCI machine learning repository which gives the dataset for 3 penguin species in the islands of Palmer Archipelago, Antarctica. It has some basic measurements on the penguins of the 3 species.

library(reshape2)
library(ggplot2)
library(dplyr)
library(ggh4x)
library(GGally)
library(naivebayes)
library(caret)
library(nnet)

# Data is available in the palmer penguins package in R
library(palmerpenguins)


# Data processing
data0 <- penguins
head(data0)
# A tibble: 6 × 8
  species island    bill_length_mm bill_depth_mm flipper_length_mm body_mass_g
  <fct>   <fct>              <dbl>         <dbl>             <int>       <int>
1 Adelie  Torgersen           39.1          18.7               181        3750
2 Adelie  Torgersen           39.5          17.4               186        3800
3 Adelie  Torgersen           40.3          18                 195        3250
4 Adelie  Torgersen           NA            NA                  NA          NA
5 Adelie  Torgersen           36.7          19.3               193        3450
6 Adelie  Torgersen           39.3          20.6               190        3650
# ℹ 2 more variables: sex <fct>, year <int>
data0 |> str()
tibble [344 × 8] (S3: tbl_df/tbl/data.frame)
 $ species          : Factor w/ 3 levels "Adelie","Chinstrap",..: 1 1 1 1 1 1 1 1 1 1 ...
 $ island           : Factor w/ 3 levels "Biscoe","Dream",..: 3 3 3 3 3 3 3 3 3 3 ...
 $ bill_length_mm   : num [1:344] 39.1 39.5 40.3 NA 36.7 39.3 38.9 39.2 34.1 42 ...
 $ bill_depth_mm    : num [1:344] 18.7 17.4 18 NA 19.3 20.6 17.8 19.6 18.1 20.2 ...
 $ flipper_length_mm: int [1:344] 181 186 195 NA 193 190 181 195 193 190 ...
 $ body_mass_g      : int [1:344] 3750 3800 3250 NA 3450 3650 3625 4675 3475 4250 ...
 $ sex              : Factor w/ 2 levels "female","male": 2 1 1 NA 1 2 1 2 NA NA ...
 $ year             : int [1:344] 2007 2007 2007 2007 2007 2007 2007 2007 2007 2007 ...
# Check the rows which do not have any entries
sum(is.na(data0)) # 19 NA values
[1] 19
# exclude the rows which has NA in them
data00 <- na.omit(data0)
# Pairs plot between the explanatory variables to 
# check correlation between each pair of the variables
ggpairs(data00[,-c(2,8)])

It is not unexpected to see multicollinearity in the data for the continuous variables since they are body measurements for the penguins.

# Histogram based on species
melted_data <- melt(data00[,c(1,3,4,5,6)], id="species")

# Plot the histogram of all the variables
ggplot(melted_data,aes(value))+
  geom_histogram(aes(),bins = 30)+
  facet_grid2(species~variable, scales="free")+theme_bw()

# Histogram based on sex
melted_data2 <- melt(data00[,c(3,4,5,6,7)], id="sex")

# Plot the histogram of all the variables
ggplot(melted_data2,aes(value))+
  geom_histogram(aes(),bins = 30)+
  facet_grid2(sex~variable, scales="free")+theme_bw()

There is a distinct difference in the histogram of all the variables based on the species and sex. We will look to build a naive Bayes classification model to identify the species of penguins. First let us split the data into training and testing set.

# Select variables to be used in the model
data <- data00 %>% dplyr::select(species,bill_length_mm,bill_depth_mm,
                                 flipper_length_mm,body_mass_g)

data1 <- subset(data, species == "Adelie")
data2 <- subset(data, species == "Chinstrap")
data3 <- subset(data, species == "Gentoo")

# split the data into training (70%) and testing (30%) data
seed <- 33
set.seed(seed)
ind1 <- sample(floor(0.7*nrow(data1)),
              replace = FALSE)

ind2 <- sample(floor(0.7*nrow(data2)),
               replace = FALSE)

ind3 <- sample(floor(0.7*nrow(data3)),
               replace = FALSE)


# Training dataset
data_train <- rbind(data1[ind1,],data2[ind2,],data3[ind3,])
data_train |> count(species)
# A tibble: 3 × 2
  species       n
  <fct>     <int>
1 Adelie      102
2 Chinstrap    47
3 Gentoo       83
# Testing dataset
data_test <- rbind(data1[-ind1,],data2[-ind2,],data3[-ind3,])
data_test |> count(species)
# A tibble: 3 × 2
  species       n
  <fct>     <int>
1 Adelie       44
2 Chinstrap    21
3 Gentoo       36

We now build a multinomial logistic regression model to identify the species of Penguins. A multinomial logistic regression model is used when we have more than 2 nominal classes in a categorical variables.

model2 <- multinom(species ~ ., data = data_train) 
# weights:  18 (10 variable)
initial  value 254.878051 
iter  10 value 18.171584
iter  20 value 3.530079
iter  30 value 1.576129
iter  40 value 0.007276
iter  50 value 0.000719
iter  60 value 0.000656
iter  70 value 0.000216
iter  80 value 0.000214
iter  90 value 0.000195
iter 100 value 0.000192
final  value 0.000192 
stopped after 100 iterations
model2 |> summary()
Call:
multinom(formula = species ~ ., data = data_train)

Coefficients:
          (Intercept) bill_length_mm bill_depth_mm flipper_length_mm
Chinstrap   -4.627397       50.77590     -53.80151        -4.4291595
Gentoo      -3.282477       26.93981     -66.79718        -0.9805378
          body_mass_g
Chinstrap -0.10625251
Gentoo     0.04238281

Std. Errors:
           (Intercept) bill_length_mm bill_depth_mm flipper_length_mm
Chinstrap 0.0324946625    0.792550472   0.351600282        4.77472729
Gentoo    0.0001474703    0.006326474   0.002595476        0.02890417
          body_mass_g
Chinstrap   0.2441534
Gentoo      0.6931102

Residual Deviance: 0.00038416 
AIC: 20.00038 
y_pred <- predict(model2, newdata = data_test)
conf_table <- table(data_test$species, y_pred)
confusionMatrix(conf_table)
Confusion Matrix and Statistics

           y_pred
            Adelie Chinstrap Gentoo
  Adelie        44         0      0
  Chinstrap      1        20      0
  Gentoo         0         0     36

Overall Statistics
                                          
               Accuracy : 0.9901          
                 95% CI : (0.9461, 0.9997)
    No Information Rate : 0.4455          
    P-Value [Acc > NIR] : < 2.2e-16       
                                          
                  Kappa : 0.9845          
                                          
 Mcnemar's Test P-Value : NA              

Statistics by Class:

                     Class: Adelie Class: Chinstrap Class: Gentoo
Sensitivity                 0.9778           1.0000        1.0000
Specificity                 1.0000           0.9877        1.0000
Pos Pred Value              1.0000           0.9524        1.0000
Neg Pred Value              0.9825           1.0000        1.0000
Prevalence                  0.4455           0.1980        0.3564
Detection Rate              0.4356           0.1980        0.3564
Detection Prevalence        0.4356           0.2079        0.3564
Balanced Accuracy           0.9889           0.9938        1.0000
# Misclassification
1 - sum(diag(conf_table)) / sum(conf_table)
[1] 0.00990099

The misclassification rate of the model on test dataset is about 1% which is not bad. In fact we misclassified the species of only one penguin.