基于CSV数据训练测试模型

基于语音进行情绪识别

数据类型为csv文件，其中包含847个特征列。类别有三类：happy,surprised,angry。
数据集可以在kaggle上下载
链接地址：https://www.kaggle.com/ashishbansal23/emotion-recognition

基本包的导入

import numpy as np
import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.model_selection import GridSearchCV
from sklearn.svm import SVC
from sklearn.ensemble import RandomForestClassifier
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import accuracy_score, classification_report, confusion_matrix

这里使用了三个机器学习模型：SVC，RandomForestClassifier，LinearRegression进行实验。GridSearchCV用来循环查找最好的参数。

读csv文件

将name和emotion列删除，其他的数据作为特征X。emotion列作为标签y。

df = pd.read_csv('E:/emotion_recognition/project1/ANAD_Normalized.csv')
X=df.drop(['name', 'Emotion '], axis=1) # features
y=df['Emotion ']     # labels
X_train, X_test, y_train, y_test = train_test_split(X,y, test_size=0.25, random_state=1) # 划分数据集

选择模型

1. SVC模型

grid={'C':[1,5,50],
     'gamma':[0.1,1,10,100]}
model=GridSearchCV(SVC(), grid)
model.fit(X_train, y_train)
pred=model.predict(X_test)
print(classification_report(y_test, pred))

运行结果：

                precision    recall  f1-score   support

       angry       0.97      0.98      0.97       177
       happy       0.94      0.97      0.95       134
   surprised       0.96      0.77      0.86        35

    accuracy                           0.95       346
   macro avg       0.96      0.91      0.93       346
weighted avg       0.95      0.95      0.95       346

利用model.best_params_可以输出最佳的参数

{'C': 5, 'gamma': 0.1}

2. RandomForestClassifier模型

grid={'n_estimators':[10,50,100,300]}
model=GridSearchCV(RandomForestClassifier(), grid)
model.fit(X_train,y_train)
print(model.best_params_)

打印最佳参数

{'n_estimators': 100}

模型评估

pred=model.predict(X_test)
print(classification_report(y_test, pred2))

打印分类评估指标报告

                precision    recall  f1-score   support

       angry       0.94      0.99      0.97       177
       happy       0.95      0.96      0.95       134
   surprised       0.92      0.63      0.75        35

    accuracy                           0.94       346
   macro avg       0.94      0.86      0.89       346
weighted avg       0.94      0.94      0.94       346

3. LogisticRegression模型

model = LogisticRegression(solver='liblinear',multi_class='auto')
model.fit(X_train,y_train)
pred=model.predict(X_test)
# accuracy_score(y_test, pred)
print(classification_report(y_test, pred))

运行结果

                precision    recall  f1-score   support

       angry       0.97      0.95      0.96       177
       happy       0.93      0.98      0.95       134
   surprised       0.87      0.77      0.82        35

    accuracy                           0.95       346
   macro avg       0.92      0.90      0.91       346
weighted avg       0.94      0.95      0.94       346