Seminar R – Mushroom Classification

Student information

Name	M_Nr
Malte Gunnar Grube	221200075
Maximilian Runge	220200831
Pascal Pischel	221201473
Truong An Nguyen	221200154

---

1. Introducing words

1-a) Background and motivation

Mushrooms are widely consumed in various countries due to their culinary appeal. However, the presence of poisonous mushrooms poses a significant risk to human health and safety. Therefore, it is particularly important to be able to distinguish between edible and poisonous mushrooms. This task can be very challenging, as there are many different types of mushrooms species. In this project, our objective is to address this challenge by using a comprehensive dataset to train and employ a machine learning model capable of accurately distinguish between edible and poisonous mushrooms.

1-b) Aim of the project

For this project, we aim to enhance the reliability and efficiency of mushroom classification, ultimately promoting public health and minimizing the risks associated with mushroom consumption. To accomplish this, we will make use of a dataset that contains relevant information about mushrooms. We will ultimately train a model, to effectively analyze and interpret the provided mushroom data, enabling it to classify mushrooms into edible and poisonous.

---

2. Dataset and data preparation

2-a) Dataset description

The dataset we are going to deal with is the Mushroom Classification dataset from Kaggle. It contains information about mushrooms and whether they are edible or poisonous on 8124 samples. The dataset contains 23 columns, of which 22 are features and 1 is the target variable (edible or poisonous).

2-b) Dataset columns

Introducing, we will take a closer look at the columns of the dataset.

Read dataset

Let’s take a look at how to read the dataset. The dataset is available as a csv file. We can read it with the following command:

mush <- read.csv("mushrooms.csv")

Classification of the columns

To understand what letter stands for what, we will take a look at the classification of the columns:

Attribute information

Attribute	Classes
cap-shape	bell=b, conical=c, convex=x, flat=f, knobbed=k, sunken=s
cap-surface	fibrous=f, grooves=g, scaly=y, smooth=s
cap-color	brown=n, buff=b, cinnamon=c, gray=g, green=r, pink=p, purple=u, red=e, white=w, yellow=y
bruises	bruises=t, no=f
odor	almond=a, anise=l, creosote=c, fishy=y, foul=f, musty=m, none=n, pungent=p, spicy=s
gill-attachment	attached=a, descending=d, free=f, notched=n
gill-spacing	close=c, crowded=w, distant=d
gill-size	broad=b, narrow=n
gill-color	black=k, brown=n, buff=b, chocolate=h, gray=g, green=r, orange=o, pink=p, purple=u, red=e, white=w, yellow=y
stalk-shape	enlarging=e, tapering=t
stalk-root	bulbous=b, club=c, cup=u, equal=e, rhizomorphs=z, rooted=r, missing=?
stalk-surface-above-ring	fibrous=f, scaly=y, silky=k, smooth=s
stalk-surface-below-ring	fibrous=f, scaly=y, silky=k, smooth=s
stalk-color-above-ring	brown=n, buff=b, cinnamon=c, gray=g, orange=o, pink=p, red=e, white=w, yellow=y
stalk-color-below-ring	brown=n, buff=b, cinnamon=c, gray=g, orange=o, pink=p, red=e, white=w, yellow=y
veil-type	partial=p, universal=u
veil-color	brown=n, orange=o, white=w, yellow=y
ring-number	none=n, one=o, two=t
ring-type	cobwebby=c, evanescent=e, flaring=f, large=l, none=n, pendant=p, sheathing=s, zone=z
spore-print-color	black=k, brown=n, buff=b, chocolate=h, green=r, orange=o, purple=u, white=w, yellow=y
population	abundant=a, clustered=c, numerous=n, scattered=s, several=v, solitary=y
habitat	grasses=g, leaves=l, meadows=m, paths=p, urban=u, waste=w, woods=d

Check null columns

Now we will check if there are any null columns in the dataset:

colSums(is.na(mush))

                   class                cap.shape              cap.surface 
                       0                        0                        0 
               cap.color                  bruises                     odor 
                       0                        0                        0 
         gill.attachment             gill.spacing                gill.size 
                       0                        0                        0 
              gill.color              stalk.shape               stalk.root 
                       0                        0                        0 
stalk.surface.above.ring stalk.surface.below.ring   stalk.color.above.ring 
                       0                        0                        0 
  stalk.color.below.ring                veil.type               veil.color 
                       0                        0                        0 
             ring.number                ring.type        spore.print.color 
                       0                        0                        0 
              population                  habitat 
                       0                        0 

Structure

Let’s check the structure of the dataset:

str(mush)

'data.frame':   8124 obs. of  23 variables:
 $ class                   : chr  "p" "e" "e" "p" ...
 $ cap.shape               : chr  "x" "x" "b" "x" ...
 $ cap.surface             : chr  "s" "s" "s" "y" ...
 $ cap.color               : chr  "n" "y" "w" "w" ...
 $ bruises                 : chr  "t" "t" "t" "t" ...
 $ odor                    : chr  "p" "a" "l" "p" ...
 $ gill.attachment         : chr  "f" "f" "f" "f" ...
 $ gill.spacing            : chr  "c" "c" "c" "c" ...
 $ gill.size               : chr  "n" "b" "b" "n" ...
 $ gill.color              : chr  "k" "k" "n" "n" ...
 $ stalk.shape             : chr  "e" "e" "e" "e" ...
 $ stalk.root              : chr  "e" "c" "c" "e" ...
 $ stalk.surface.above.ring: chr  "s" "s" "s" "s" ...
 $ stalk.surface.below.ring: chr  "s" "s" "s" "s" ...
 $ stalk.color.above.ring  : chr  "w" "w" "w" "w" ...
 $ stalk.color.below.ring  : chr  "w" "w" "w" "w" ...
 $ veil.type               : chr  "p" "p" "p" "p" ...
 $ veil.color              : chr  "w" "w" "w" "w" ...
 $ ring.number             : chr  "o" "o" "o" "o" ...
 $ ring.type               : chr  "p" "p" "p" "p" ...
 $ spore.print.color       : chr  "k" "n" "n" "k" ...
 $ population              : chr  "s" "n" "n" "s" ...
 $ habitat                 : chr  "u" "g" "m" "u" ...

2-c) Data preparation

In this section, we will prepare the data for the model.

Remove veil-type

After this, we will remove the column veil-type, as it only contains one value called p:

mush <- data.frame(mush[, -17])

Convert categorical variables

Finally, we will convert the variables into factors – because all features are categorical. Here is a preview of the converted variables:

mush <- as.data.frame(lapply(mush, as.factor))

  Converted variables

class	cap.shape	cap.surface	cap.color	bruises	odor	gill.attachment	gill.spacing	gill.size	gill.color	stalk.shape	stalk.root	stalk.surface.above.ring	stalk.surface.below.ring	stalk.color.above.ring	stalk.color.below.ring	veil.color	ring.number	ring.type	spore.print.color	population	habitat
p	x	s	n	t	p	f	c	n	k	e	e	s	s	w	w	w	o	p	k	s	u
e	x	s	y	t	a	f	c	b	k	e	c	s	s	w	w	w	o	p	n	n	g
e	b	s	w	t	l	f	c	b	n	e	c	s	s	w	w	w	o	p	n	n	m
p	x	y	w	t	p	f	c	n	n	e	e	s	s	w	w	w	o	p	k	s	u
e	x	s	g	f	n	f	w	b	k	t	e	s	s	w	w	w	o	e	n	a	g
e	x	y	y	t	a	f	c	b	n	e	c	s	s	w	w	w	o	p	k	n	g

2-d) Data visualization

Poisonous-edible distribution

Here you can see the distribution of the both classes:

ggplot(data = mush, aes(x = class, fill = class)) +
  geom_bar(alpha = 0.7) +
  labs(title = "class", x = NULL, y = NULL) +
  scale_fill_manual(values = c("#33cc33", "#4b0092")) +
  theme_bw() +
  theme(
    plot.title = element_text(hjust = 0.5, size = 11, face = "bold"),
    legend.position = "none"
  )

e / p results

class	n
edible	4208
poisonous	3916

---

Feature distributions

Let’s take a look at the distributions of all the features:

feature_names <- names(mush)[-1]

feature_plots <- lapply(feature_names, function(feature) {
  ggplot(data = mush, aes(x = get(feature), fill = class)) +
    geom_bar(alpha = 0.7) +
    labs(title = paste(feature), x = NULL, y = NULL) +
    scale_fill_manual(
      values = c("#33cc33", "#4b0092"),
      labels = c("edible", "poisonous")
    ) +
    theme_bw() +
    theme(
      plot.title = element_text(hjust = 0.5, size = 11, face = "bold"),
      legend.position = "top"
    ) +
    guides(fill = guide_legend(title = "class", nrow = 2))
})

legend <- get_legend(feature_plots[[1]])

for (i in seq_along(feature_plots)) {
  feature_plots[[i]] <- feature_plots[[i]] + theme(legend.position = "none")
}

plot_grid(plotlist = feature_plots, ncol = 2, nrow = 11, legend = legend)

---

Edible features

In these plots, we can see that certain features are important to classify a mushroom as edible.
If a test mushroom has one of these features, it is more likely to be this class:

Edible features

Attribute	Description
odor	almond=a, anise=l, none=n
stalk.color.below.ring	red=e, orange=o, gray=g
stalk.color.above.ring	red=e, orange=o, gray=g
spore.print.color	yellow=y, purple=u, orange=o, buff=b
population	abundant=a, numerous=n
veil.color	orange=o, brown=n
cap.color	red=e, purple=u
gill.color	gray=g, orange=o
ring.type	flaring=f
cap.shape	sunken=s

---

Poisonous features

In these plots, we can see that certain features are important to classify a mushroom as poisonous.
If a test mushroom has one of these features, it is more likely to be this class:

Poisonous features

Attribute	Description
odor	pungent=p, foul=f, creosote=c, fishy=y, spicy=s, musty=m
stalk.color.below.ring	buff=b, yellow=y, cinnamon=c
ring.number	none=n
gill.color	buff=b, green=r
spore.print.color	green=r
cap.shape	conical=c
veil.color	yellow=y
stalk.color.above.ring	buff=b, yellow=y, cinnamon=c
ring.type	large=l
cap.surface	grooves=g

---

3. Machine learning

3-a) Model preparation

Now we will split the data into training and test data. We will use the training data to train the model and the test data to test the model.

set.seed(753)
train_index <- createDataPartition(mush$class,
  p = .8, list = FALSE, times = 1
)
train <- mush[train_index, ]
test <- mush[-train_index, ]

3-b) Training models

In this section, we will train several models and compare them – sped up via parallel processing.

GLM

Let’s take a look at the Logistic Regression (glm) model:

cl <- makeCluster(detectCores() - 1)
registerDoParallel(cl)
start_time <- Sys.time()
model_glm <- train(class ~ ., data = train, method = "glm")
stopCluster(cl)
end_time <- Sys.time()
end_time - start_time

Time difference of 17.80068 secs

SVM

Let’s take a look at the Support Vector Machines (linear) model:

cl <- makeCluster(detectCores() - 1)
registerDoParallel(cl)
start_time <- Sys.time()
model_svm <- train(class ~ ., data = train, method = "svmLinear")
stopCluster(cl)
end_time <- Sys.time()
end_time - start_time

Time difference of 8.636677 secs

RF

Let’s take a look at the Random Forest model:

cl <- makeCluster(detectCores() - 1)
registerDoParallel(cl)
start_time <- Sys.time()
model_rf <- train(class ~ ., data = train, method = "rf")
stopCluster(cl)
end_time <- Sys.time()
end_time - start_time

Time difference of 2.940914 mins

KNN

Let’s take a look at the K-Nearest Neighbors model:

cl <- makeCluster(detectCores() - 1)
registerDoParallel(cl)
start_time <- Sys.time()
model_knn <- train(class ~ ., data = train, method = "knn")
stopCluster(cl)
end_time <- Sys.time()
end_time - start_time

Time difference of 49.82261 secs

GBM

Let’s take a look at the Gradient Boosting Machine model:

cl <- makeCluster(detectCores() - 1)
registerDoParallel(cl)
start_time <- Sys.time()
model_gbm <- train(class ~ ., data = train, method = "gbm", verbose = FALSE)
stopCluster(cl)
end_time <- Sys.time()
end_time - start_time

Time difference of 24.34671 secs

3-c) Confusion matrices

GLM

Let’s take a look at the confusion matrix of the Logistic Regression (glm) model:

confusion_glm <- confusionMatrix(predict(model_glm, test), test$class)
confusion_glm

Confusion Matrix and Statistics

          Reference
Prediction   e   p
         e 841   0
         p   0 783
                                     
               Accuracy : 1          
                 95% CI : (0.9977, 1)
    No Information Rate : 0.5179     
    P-Value [Acc > NIR] : < 2.2e-16  
                                     
                  Kappa : 1          
                                     
 Mcnemar's Test P-Value : NA         
                                     
            Sensitivity : 1.0000     
            Specificity : 1.0000     
         Pos Pred Value : 1.0000     
         Neg Pred Value : 1.0000     
             Prevalence : 0.5179     
         Detection Rate : 0.5179     
   Detection Prevalence : 0.5179     
      Balanced Accuracy : 1.0000     
                                     
       'Positive' Class : e          
                                     

SVM

Let’s take a look at the confusion matrix of the Support Vector Machines (linear) model:

confusion_svm <- confusionMatrix(predict(model_svm, test), test$class)
confusion_svm

Confusion Matrix and Statistics

          Reference
Prediction   e   p
         e 841   0
         p   0 783
                                     
               Accuracy : 1          
                 95% CI : (0.9977, 1)
    No Information Rate : 0.5179     
    P-Value [Acc > NIR] : < 2.2e-16  
                                     
                  Kappa : 1          
                                     
 Mcnemar's Test P-Value : NA         
                                     
            Sensitivity : 1.0000     
            Specificity : 1.0000     
         Pos Pred Value : 1.0000     
         Neg Pred Value : 1.0000     
             Prevalence : 0.5179     
         Detection Rate : 0.5179     
   Detection Prevalence : 0.5179     
      Balanced Accuracy : 1.0000     
                                     
       'Positive' Class : e          
                                     

RF

Let’s take a look at the confusion matrix of the Random Forest model:

confusion_rf <- confusionMatrix(predict(model_rf, test), test$class)
confusion_rf

Confusion Matrix and Statistics

          Reference
Prediction   e   p
         e 841   0
         p   0 783
                                     
               Accuracy : 1          
                 95% CI : (0.9977, 1)
    No Information Rate : 0.5179     
    P-Value [Acc > NIR] : < 2.2e-16  
                                     
                  Kappa : 1          
                                     
 Mcnemar's Test P-Value : NA         
                                     
            Sensitivity : 1.0000     
            Specificity : 1.0000     
         Pos Pred Value : 1.0000     
         Neg Pred Value : 1.0000     
             Prevalence : 0.5179     
         Detection Rate : 0.5179     
   Detection Prevalence : 0.5179     
      Balanced Accuracy : 1.0000     
                                     
       'Positive' Class : e          
                                     

KNN

Let’s take a look at the confusion matrix of the K-Nearest Neighbors model:

confusion_knn <- confusionMatrix(predict(model_knn, test), test$class)
confusion_knn

Confusion Matrix and Statistics

          Reference
Prediction   e   p
         e 841   0
         p   0 783
                                     
               Accuracy : 1          
                 95% CI : (0.9977, 1)
    No Information Rate : 0.5179     
    P-Value [Acc > NIR] : < 2.2e-16  
                                     
                  Kappa : 1          
                                     
 Mcnemar's Test P-Value : NA         
                                     
            Sensitivity : 1.0000     
            Specificity : 1.0000     
         Pos Pred Value : 1.0000     
         Neg Pred Value : 1.0000     
             Prevalence : 0.5179     
         Detection Rate : 0.5179     
   Detection Prevalence : 0.5179     
      Balanced Accuracy : 1.0000     
                                     
       'Positive' Class : e          
                                     

GBM

Let’s take a look at the confusion matrix of the Gradient Boosting Machine model:

confusion_gbm <- confusionMatrix(predict(model_gbm, test), test$class)
confusion_gbm

Confusion Matrix and Statistics

          Reference
Prediction   e   p
         e 841   0
         p   0 783
                                     
               Accuracy : 1          
                 95% CI : (0.9977, 1)
    No Information Rate : 0.5179     
    P-Value [Acc > NIR] : < 2.2e-16  
                                     
                  Kappa : 1          
                                     
 Mcnemar's Test P-Value : NA         
                                     
            Sensitivity : 1.0000     
            Specificity : 1.0000     
         Pos Pred Value : 1.0000     
         Neg Pred Value : 1.0000     
             Prevalence : 0.5179     
         Detection Rate : 0.5179     
   Detection Prevalence : 0.5179     
      Balanced Accuracy : 1.0000     
                                     
       'Positive' Class : e          
                                     

3-d-I) Importance plot

One of the most accurate models is RandomForest. With RandomForest the function varImp() shows the permutation importance. As shown in the importance plot of the model, the odor none is the most significant indicator for an edible mushroom.

var_importance <- varImp(model_rf$finalModel)
var_df <- as.data.frame(var_importance)
sorted_vars <- var_df %>%
  arrange(desc(Overall)) %>%
  head(20)

total <- sum(sorted_vars$Overall)
sorted_vars$Overall <- (sorted_vars$Overall / total) * 100

x_labels <- str_sub(row.names(sorted_vars), end = -2) %>%
  paste0("_", str_sub(row.names(sorted_vars), -1))

ggplot(sorted_vars, aes(
  x = reorder(row.names(sorted_vars), -Overall),
  y = Overall
)) +
  geom_bar(stat = "identity", alpha = 0.6, fill = "#00303F") +
  labs(x = NULL, y = NULL, title = "Variable vs. Importance (in %)") +
  coord_flip() +
  theme_bw() +
  theme(
    legend.position = "none",
    plot.title = element_text(hjust = 0.5, face = "bold")
  ) +
  scale_x_discrete(labels = x_labels)

3-d-II) Relation plots

Here are some plots that show the relationship between some of the variables and the class of the mushrooms.

Gill size vs. Odor

ggplot(mush, aes(x = gill.size, y = odor)) +
  geom_jitter(aes(color = class, alpha = class),
    position = position_jitter(width = 0.2)
  ) +
  scale_color_manual(values = c("e" = "#33cc33", "p" = "#4b0092")) +
  scale_alpha_manual(values = c("e" = 0.4, "p" = 0.6)) +
  scale_size(guide = "none") +
  labs(title = "Gill size vs. Odor") +
  theme_bw() +
  theme(
    plot.title = element_text(
      hjust = 0.5, face = "bold",
      margin = ggplot2::margin(0, 0, 5, 0)
    ),
    axis.title.x = element_text(margin = ggplot2::margin(7, 0, 0, 0)),
    axis.title.y = element_text(margin = ggplot2::margin(0, 15, 0, 0))
  )

Spore print color vs. Gill color

ggplot(mush, aes(x = spore.print.color, y = gill.color)) +
  geom_jitter(aes(color = class, alpha = class),
    position = position_jitter(width = 0.2)
  ) +
  scale_color_manual(values = c("e" = "#33cc33", "p" = "#4b0092")) +
  scale_alpha_manual(values = c("e" = 0.4, "p" = 0.6)) +
  scale_size(guide = "none") +
  labs(title = "Spore print color vs. Gill color") +
  theme_bw() +
  theme(
    plot.title = element_text(
      hjust = 0.5, face = "bold",
      margin = ggplot2::margin(0, 0, 5, 0)
    ),
    axis.title.x = element_text(margin = ggplot2::margin(7, 0, 0, 0)),
    axis.title.y = element_text(margin = ggplot2::margin(0, 15, 0, 0))
  )

Habitat vs. Population

ggplot(mush, aes(x = habitat, y = population)) +
  geom_jitter(aes(color = class, alpha = class),
    position = position_jitter(width = 0.2)
  ) +
  scale_color_manual(values = c("e" = "#33cc33", "p" = "#4b0092")) +
  scale_alpha_manual(values = c("e" = 0.4, "p" = 0.6)) +
  scale_size(guide = "none") +
  labs(title = "Habitat vs. Population") +
  theme_bw() +
  theme(
    plot.title = element_text(
      hjust = 0.5, face = "bold",
      margin = ggplot2::margin(0, 0, 5, 0)
    ),
    axis.title.x = element_text(margin = ggplot2::margin(7, 0, 0, 0)),
    axis.title.y = element_text(margin = ggplot2::margin(0, 15, 0, 0))
  )

Cap shape vs. Stalk color below ring

ggplot(mush, aes(x = cap.shape, y = stalk.color.below.ring)) +
  geom_jitter(aes(color = class, alpha = class),
    position = position_jitter(width = 0.2)
  ) +
  scale_color_manual(values = c("e" = "#33cc33", "p" = "#4b0092")) +
  scale_alpha_manual(values = c("e" = 0.4, "p" = 0.6)) +
  scale_size(guide = "none") +
  labs(title = "Cap shape vs. Stalk color below ring") +
  theme_bw() +
  theme(
    plot.title = element_text(
      hjust = 0.5, face = "bold",
      margin = ggplot2::margin(0, 0, 5, 0)
    ),
    axis.title.x = element_text(margin = ggplot2::margin(7, 0, 0, 0)),
    axis.title.y = element_text(margin = ggplot2::margin(0, 15, 0, 0))
  )

3-e) Decision trees

Finishing the document, we are illustrating various decision trees. The first one is a general tree. Then, we are going to remove the odor variable and see how the tree changes. The next tree does not provide properties that are hard to tell. The last one just provides properties according to the top view of a mushroom.

mush_fac <- as.data.frame(lapply(mush, as.factor))
loss_matrix <- matrix(c(0, 1, 4, 0), byrow = TRUE, nrow = 2)

General tree

tree1 <- rpart(class ~ .,
  minbucket = 1,
  method = "class",
  parms = list(loss = loss_matrix),
  data = mush_fac
)

fancyRpartPlot(tree1,
  palettes = c("Greens", "Purples"),
  caption = NULL, cex = 0.8,
  main = "General tree"
)

---

Without Odor

mush_fac_no_odor <- mush_fac
mush_fac_no_odor$odor <- NULL

tree2 <- rpart(class ~ .,
  minbucket = 2,
  method = "class",
  parms = list(loss = loss_matrix),
  data = mush_fac_no_odor
)

fancyRpartPlot(tree2,
  palettes = c("Greens", "Purples"),
  caption = NULL, cex = 0.8,
  main = "Without Odor"
)

---

Without hard to tell properties

mush_fac_hard <- mush_fac %>%
  select(
    -odor, -spore.print.color, -gill.size, -bruises, -ring.type,
    -stalk.surface.above.ring, -stalk.surface.below.ring, -stalk.root
  ) %>%
  mutate_at(vars(
    gill.color, cap.color, stalk.color.above.ring,
    stalk.color.below.ring
  ), ~ if_else(. %in% c("b", "h", "c"), "n", .))

tree3 <- rpart(class ~ .,
  minbucket = 200,
  method = "class",
  parms = list(loss = loss_matrix),
  data = mush_fac_hard
)

fancyRpartPlot(tree3,
  palettes = c("Greens", "Purples"),
  caption = NULL,
  main = "Without hard to tell properties"
)

---

Top view only

mush_fac_top <- subset(mush_fac, select = -c(
  gill.size,
  gill.color, gill.attachment, gill.spacing,
  stalk.shape, stalk.root, stalk.surface.above.ring,
  stalk.surface.below.ring, stalk.color.above.ring,
  stalk.color.below.ring, spore.print.color
))

tree4 <- rpart(class ~ .,
  minbucket = 1,
  method = "class",
  parms = list(loss = loss_matrix),
  data = mush_fac_top
)

fancyRpartPlot(tree4,
  palettes = c("Greens", "Purples"),
  caption = NULL, cex = 0.8,
  main = "Top view only"
)

---

4. Real mushrooms

In this section, we are going to show some real mushrooms. In addition to that, we are going to provide trees for classfying them.

Shield Dapperling [1]

fancyRpartPlot(tree3,
  palettes = c("Greens", "Purples"),
  caption = NULL, main = NULL
)

---

Deadly Dapperling [2]

fancyRpartPlot(tree3,
  palettes = c("Greens", "Purples"),
  caption = NULL, main = NULL
)

---

Medusa Mushroom [3]

fancyRpartPlot(tree3,
  palettes = c("Greens", "Purples"),
  caption = NULL, main = NULL
)

---

Inky Mushroom [4]

fancyRpartPlot(tree3,
  palettes = c("Greens", "Purples"),
  caption = NULL, main = NULL
)

---

5. Conclusion

As we delved into the analysis of the mushroom dataset, exploring its various features and their importance, we were able to train several models and compare them. With that we gained valuable insights into the characteristics that differentiate edible and poisonous mushrooms. Through our efforts, we successfully developed decision trees that proved effective in classifying real mushrooms based on the knowledge obtained from the dataset. In conclusion, this project serves as a foundation for future mushroom classification and related studies. By utilizing the power of machine learning, we can continue to make strides in understanding and utilizing the world of mushrooms for the betterment of society.

---

References

[1] iNaturalist: Shield Dapperling. URL: https://www.inaturalist.org/observations/156623219. Accessed: July 20, 2023
[2] iNaturalist: Deadly Dapperling. URL: https://www.inaturalist.org/observations/70759242. Accessed: July 20, 2023.
[3] iNaturalist: Medusa Mushroom. URL: https://www.inaturalist.org/observations/91868779. Accessed: July 20, 2023.
[4] iNaturalist: Inky Mushroom. URL: https://www.inaturalist.org/observations/170844438. Accessed: July 20, 2023.