emotion-detection-nlp
January 6, 2024
[38]: #import necessary library
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
import spacy
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression
from sklearn.neighbors import KNeighborsClassifier
from sklearn.naive_bayes import MultinomialNB
from sklearn.ensemble import RandomForestClassifier
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.pipeline import Pipeline
from sklearn.metrics import classification_report
from sklearn.metrics import confusion_matrix
import warnings
warnings.filterwarnings('ignore')
1 Load The Dataset
[39]: df = pd.read_csv("/kaggle/input/emotiondata/train.txt",sep=";
↪",names=["Description","Emotion"])
df.head()
[39]: Description Emotion
0 i didnt feel humiliated sadness
1 i can go from feeling so hopeless to so damned… sadness
2 im grabbing a minute to post i feel greedy wrong anger
3 i am ever feeling nostalgic about the fireplac… love
4 i am feeling grouchy anger
[40]: df['Emotion'].value_counts()
1
[40]: Emotion
joy 5362
sadness 4666
anger 2159
fear 1937
love 1304
surprise 572
Name: count, dtype: int64
1.1 Map emotions to numerical values
[41]: df['Emotion_num'] = df['Emotion'].map({
'joy' : 0,
'sadness': 1,
'anger': 2,
'fear': 3,
'love': 4,
'surprise':5
})
df.head()
[41]: Description Emotion Emotion_num
0 i didnt feel humiliated sadness 1
1 i can go from feeling so hopeless to so damned… sadness 1
2 im grabbing a minute to post i feel greedy wrong anger 2
3 i am ever feeling nostalgic about the fireplac… love 4
4 i am feeling grouchy anger 2
[42]: #Bar plot for emotion distribution
plt.figure(figsize=(8, 6))
sns.countplot(x='Emotion', data=df, order=df['Emotion'].value_counts().index,␣
↪palette="husl")
plt.title('Distribution of Emotions')
plt.show()
2
[43]: # Histogram plot for emotion distribution with KDE
plt.figure(figsize=(10, 6))
sns.histplot(x='Emotion', data=df, kde=True, palette='Set2', element='bars',␣
↪stat='count',
common_norm=False)
plt.title('Distribution of Emotions')
plt.xticks(rotation=45, ha='right')
plt.show()
3
[44]: # Load spaCy English model
nlp = spacy.load("en_core_web_sm")
# Function to preprocess text using spaCy
def preprocess(text):
doc = nlp(text)
filtered_tokens = []
for token in doc:
if token.is_stop or token.is_punct:
continue
else:
filtered_tokens.append(token.lemma_)
return " ".join(filtered_tokens)
[45]: df['processed_text'] = df["Description"].apply(preprocess)
df
[45]: Description Emotion \
0 i didnt feel humiliated sadness
1 i can go from feeling so hopeless to so damned… sadness
2 im grabbing a minute to post i feel greedy wrong anger
3 i am ever feeling nostalgic about the fireplac… love
4 i am feeling grouchy anger
4
… … …
15995 i just had a very brief time in the beanbag an… sadness
15996 i am now turning and i feel pathetic that i am… sadness
15997 i feel strong and good overall joy
15998 i feel like this was such a rude comment and i… anger
15999 i know a lot but i feel so stupid because i ca… sadness
Emotion_num processed_text
0 1 not feel humiliate
1 1 feel hopeless damned hopeful care awake
2 2 m grab minute post feel greedy wrong
3 4 feel nostalgic fireplace know property
4 2 feel grouchy
… … …
15995 1 brief time beanbag say anna feel like beat
15996 1 turn feel pathetic wait table sub teaching degree
15997 0 feel strong good overall
15998 2 feel like rude comment m glad t
15999 1 know lot feel stupid portray
[16000 rows x 4 columns]
2 Train-test split
[46]: from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(df['Description'],␣
↪df['Emotion_num'], test_size=0.2, random_state=42)
3 KNN
[47]: # KNN
knn = Pipeline([
('tfidf', TfidfVectorizer()),
('classifier', KNeighborsClassifier())
])
knn.fit(X_train, y_train)
knn_y_pred = knn.predict(X_test)
[48]: print("Classification Report:\n", classification_report(y_test, knn_y_pred))
Classification Report:
precision recall f1-score support
0 0.66 0.87 0.75 1021
1 0.73 0.82 0.77 946
5
2 0.76 0.61 0.68 427
3 0.77 0.52 0.62 397
4 0.77 0.35 0.48 296
5 0.64 0.30 0.41 113
accuracy 0.71 3200
macro avg 0.72 0.58 0.62 3200
weighted avg 0.72 0.71 0.69 3200
4 Logistic Regression
[49]: # Logistic Regression
lr = Pipeline([
('tfidf', TfidfVectorizer()),
('classifier', LogisticRegression())
])
lr.fit(X_train, y_train)
lr_y_pred = lr.predict(X_test)
[50]: print("Classification Report:\n", classification_report(y_test, lr_y_pred))
Classification Report:
precision recall f1-score support
0 0.79 0.96 0.87 1021
1 0.87 0.94 0.90 946
2 0.90 0.77 0.83 427
3 0.85 0.70 0.77 397
4 0.89 0.55 0.68 296
5 0.88 0.43 0.58 113
accuracy 0.84 3200
macro avg 0.86 0.73 0.77 3200
weighted avg 0.85 0.84 0.83 3200
5 Multinomial Naive Bayes
[51]: # Multinomial Naive Bayes
nb = Pipeline([
('tfidf', TfidfVectorizer()),
('classifier', MultinomialNB())
])
nb.fit(X_train, y_train)
nb_y_pred = nb.predict(X_test)
6
[52]: print("Classification Report:\n", classification_report(y_test, nb_y_pred))
Classification Report:
precision recall f1-score support
0 0.55 0.99 0.71 1021
1 0.69 0.90 0.78 946
2 0.91 0.17 0.29 427
3 0.93 0.11 0.19 397
4 1.00 0.01 0.01 296
5 0.00 0.00 0.00 113
accuracy 0.62 3200
macro avg 0.68 0.36 0.33 3200
weighted avg 0.71 0.62 0.52 3200
6 Random Forest
[53]: # Random Forest
rfc = Pipeline([
('tfidf', TfidfVectorizer()),
('classifier', RandomForestClassifier(random_state=42))
])
rfc.fit(X_train, y_train)
rfc_y_pred = rfc.predict(X_test)
[54]: print("Classification Report:\n", classification_report(y_test, rfc_y_pred))
Classification Report:
precision recall f1-score support
0 0.79 0.94 0.86 1021
1 0.91 0.89 0.90 946
2 0.90 0.80 0.85 427
3 0.86 0.80 0.83 397
4 0.87 0.66 0.75 296
5 0.87 0.66 0.75 113
accuracy 0.85 3200
macro avg 0.87 0.79 0.82 3200
weighted avg 0.86 0.85 0.85 3200
7
7 Confusion Matrix Heatmap for Random Forest
[56]: cm = confusion_matrix(y_test, rfc_y_pred)
cm
[56]: array([[960, 26, 4, 8, 21, 2],
[ 67, 844, 17, 11, 6, 1],
[ 43, 30, 342, 12, 0, 0],
[ 44, 13, 12, 319, 1, 8],
[ 88, 6, 6, 1, 195, 0],
[ 12, 4, 0, 22, 0, 75]])
[58]: # Plot confusion matrix
plt.figure(figsize=(10, 7))
sns.heatmap(cm, annot=True, fmt='d', cmap='coolwarm')
plt.xlabel('Prediction')
plt.ylabel('Truth')
plt.show()
