基于内容的课程推荐系统

引言

基于Coursera课程数据集，将课程名称向量化，计算与目标课程标题向量最相似的课程向量，实现基于内容的课程推荐。

代码实现

准备实验环境与数据

import numpy as np 
import pandas as pd
from statistics import harmonic_mean
from langdetect import detect
from sklearn.feature_extraction.text import TfidfVectorizer, TfidfTransformer, CountVectorizer
from sklearn.preprocessing import MinMaxScaler
from sklearn.metrics import f1_score
from sklearn.metrics.pairwise import cosine_similarity

import os
for dirname, _, filenames in os.walk('/kaggle/input'):
    for filename in filenames:
        print(os.path.join(dirname, filename))

输出：/kaggle/input/coursera-course-dataset/coursea_data.csv

展示原始数据

df = pd.read_csv('/kaggle/input/coursera-course-dataset/coursea_data.csv')
df.drop(['Unnamed: 0', 'course_organization'], axis=1, inplace=True)
df

将注册人数属性转换为数值类型

df = df[df.course_students_enrolled.str.endswith('k')]

df['course_students_enrolled'] = df['course_students_enrolled'].apply(lambda enrolled : eval(enrolled[:-1]) * 1000)
df

数据归一化

minmax_scaler = MinMaxScaler()
scaled_ratings = minmax_scaler.fit_transform(df[['course_rating','course_students_enrolled']])

df['course_rating'] = scaled_ratings[:,0]
df['course_students_enrolled'] = scaled_ratings[:,1]
df['overall_rating'] = df[['course_rating','course_students_enrolled']].apply(lambda row : harmonic_mean(row), axis=1)
df

产生推荐结果

df = df[df.course_title.apply(lambda title : detect(title) == 'en')]

vectorizer = TfidfVectorizer(stop_words='english')
vectors = vectorizer.fit_transform(df.course_title)

def recommend_by_course_title (title, recomm_count=10) : 
    title_vector = vectorizer.transform([title])
    cosine_sim = cosine_similarity(vectors, title_vector)
    idx = np.argsort(np.array(cosine_sim[:,0]))[-recomm_count:]
    sdf = df.iloc[idx].sort_values(by='overall_rating', ascending=False)
    return sdf

recommend_by_course_title('A Crash Course in Data Science')