Blood Demon Art

Download data

Back to spellbook

1. Load data

import pandas as pd
demon = pd.read_csv("demon_slayer_10.csv")
demon.head()

	ID	Character	Breathing	Moon	Form	Battle_difficulty	Effectiveness_rating	Hours_to_daybreak	Call_pillars
0	1	Tanjiro	Water	Upper	4	28	4	4	1
1	2	Inosuke	Beast	Lower	7	20	5	3	0
2	3	Zenitsu	Thunder	Upper	1	24	6	4	0
3	4	Inosuke	Beast	Lower	4	27	7	5	0
4	5	Inosuke	Beast	Upper	1	43	1	8	0

Filter required variables.

demon2 = demon.iloc[:, [1, 2, 3, 8]]
demon2.head()

	Character	Breathing	Moon	Call_pillars
0	Tanjiro	Water	Upper	1
1	Inosuke	Beast	Lower	0
2	Zenitsu	Thunder	Upper	0
3	Inosuke	Beast	Lower	0
4	Inosuke	Beast	Upper	0

2. Data partitioning

predictors = ["Character", "Breathing", "Moon"]
outcome = "Call_Pillars"

X = pd.get_dummies(demon2[predictors])
y = demon2["Call_pillars"].astype("category")

X.head()

	Character_Inosuke	Character_Nezuko	Character_Tanjiro	Character_Zenitsu	Breathing_Beast	Breathing_Blood Demon Art	Breathing_Sun	Breathing_Thunder	Breathing_Water	Moon_Lower	Moon_Upper
0	0	0	1	0	0	0	0	0	1	0	1
1	1	0	0	0	1	0	0	0	0	1	0
2	0	0	0	1	0	0	0	1	0	0	1
3	1	0	0	0	1	0	0	0	0	1	0
4	1	0	0	0	1	0	0	0	0	0	1

y.head()

0    1
1    0
2    0
3    0
4    0
Name: Call_pillars, dtype: category
Categories (2, int64): [0, 1]

Check the classes.

list(y.cat.categories)

[0, 1]

Split the data.

import sklearn
from sklearn.model_selection import train_test_split
X_train, X_valid, y_train, y_valid = train_test_split(X, y, test_size = 0.4, random_state = 666)

Check the data.

X_train.head()

	Character_Inosuke	Character_Nezuko	Character_Tanjiro	Character_Zenitsu	Breathing_Beast	Breathing_Blood Demon Art	Breathing_Sun	Breathing_Thunder	Breathing_Water	Moon_Lower	Moon_Upper
3049	1	0	0	0	1	0	0	0	0	1	0
5182	0	0	0	1	0	0	0	1	0	1	0
4913	1	0	0	0	1	0	0	0	0	0	1
4683	1	0	0	0	1	0	0	0	0	1	0
232	1	0	0	0	1	0	0	0	0	1	0

X_valid.head()

	Character_Inosuke	Character_Nezuko	Character_Tanjiro	Character_Zenitsu	Breathing_Beast	Breathing_Blood Demon Art	Breathing_Sun	Breathing_Thunder	Breathing_Water	Moon_Lower	Moon_Upper
3958	0	1	0	0	0	1	0	0	0	1	0
2967	0	1	0	0	0	1	0	0	0	1	0
746	0	0	0	1	0	0	0	1	0	1	0
1856	0	0	0	1	0	0	0	1	0	1	0
3120	0	1	0	0	0	1	0	0	0	1	0

y_valid.head()

3958    1
2967    1
746     0
1856    0
3120    1
Name: Call_pillars, dtype: category
Categories (2, int64): [0, 1]

y_valid.head()

3958    1
2967    1
746     0
1856    0
3120    1
Name: Call_pillars, dtype: category
Categories (2, int64): [0, 1]

3. Naive Bayes

import sklearn
from sklearn.naive_bayes import MultinomialNB
demon_nb_model = MultinomialNB(alpha = 0.01)
demon_nb_model

MultinomialNB(alpha=0.01)

demon_nb_model.fit(X_train, y_train)

MultinomialNB(alpha=0.01)

3.1 Probabilities

Predict probabilities, training set.

predProb_train = demon_nb_model.predict_proba(X_train)
predProb_train

array([[0.66619876, 0.33380124],
       [0.68959507, 0.31040493],
       [0.67786638, 0.32213362],
       ...,
       [0.02583048, 0.97416952],
       [0.71892676, 0.28107324],
       [0.02719663, 0.97280337]])

Predict probabilities, validation set.

predProb_valid = demon_nb_model.predict_proba(X_valid)
predProb_valid

array([[0.02583048, 0.97416952],
       [0.02583048, 0.97416952],
       [0.68959507, 0.31040493],
       ...,
       [0.02583048, 0.97416952],
       [0.66619876, 0.33380124],
       [0.67786638, 0.32213362]])

3.2 Class prediction

Predict class membership in training set.

y_train_pred = demon_nb_model.predict(X_train)
y_train_pred

array([0, 0, 0, ..., 1, 0, 1])

Predict class membership in validation set.

y_valid_pred = demon_nb_model.predict(X_valid)
y_valid_pred

array([1, 1, 0, ..., 1, 0, 0])

3.3 New record

New record

nb_new = pd.read_csv("nb_new.csv")
nb_new.head()

	Character_Inosuke	Character_Nezuko	Character_Tanjiro	Character_Zenitsu	Breathing_Beast	Breathing_Blood Demon Art	Breathing_Sun	Breathing_Thunder	Breathing_Water	Moon_Lower	Moon_Upper
0	0	1	0	0	0	1	0	0	0	0	1

predProb_nb_new = demon_nb_model.predict_proba(nb_new)
predProb_nb_new

array([[0.02719663, 0.97280337]])

nb_new_pred = demon_nb_model.predict(nb_new)
nb_new_pred

array([1])

4. Evaluation

4.1 Confusion matrix

Confusion matrix, training set.

from sklearn.metrics import confusion_matrix, accuracy_score

confusion_matrix_train = confusion_matrix(y_train, y_train_pred)
confusion_matrix_train

array([[1744,  136],
       [1244,  875]])

accuracy_train = accuracy_score(y_train, y_train_pred)
accuracy_train

0.654913728432108

from sklearn.metrics import classification_report

print(classification_report(y_train, y_train_pred))

              precision    recall  f1-score   support

           0       0.58      0.93      0.72      1880
           1       0.87      0.41      0.56      2119

    accuracy                           0.65      3999
   macro avg       0.72      0.67      0.64      3999
weighted avg       0.73      0.65      0.63      3999

Confusion matrix, validation set.

confusion_matrix_valid = confusion_matrix(y_valid, y_valid_pred)
confusion_matrix_valid

array([[1136,  105],
       [ 850,  576]])

accuracy_valid = accuracy_score(y_valid, y_valid_pred)
accuracy_valid

0.6419197600299963

print(classification_report(y_valid, y_valid_pred))

              precision    recall  f1-score   support

           0       0.57      0.92      0.70      1241
           1       0.85      0.40      0.55      1426

    accuracy                           0.64      2667
   macro avg       0.71      0.66      0.63      2667
weighted avg       0.72      0.64      0.62      2667

4.2 ROC curve for model

ROC.

from sklearn import metrics
import matplotlib.pyplot as plt

from sklearn.metrics import roc_curve

fpr1, tpr1, thresh1 = roc_curve(y_valid, predProb_valid[:,1], pos_label = 1)

import matplotlib.pyplot as plt
plt.style.use("seaborn")

plt.plot(fpr1, tpr1, linestyle = '-', color = "blue", label = "Demon Slayer")

# roc curve for tpr = fpr (random line) 
random_probs = [0 for i in range(len(y_valid))]

p_fpr, p_tpr, _ = roc_curve(y_valid, random_probs, pos_label = 1)

plt.plot(p_fpr, p_tpr, linestyle = '--', color='black')

[<matplotlib.lines.Line2D at 0x7fd990b21190>]

AUC.

from sklearn.metrics import roc_auc_score
auc = roc_auc_score(y_valid, predProb_valid[:,1])
auc

0.6774515643064849