Python《基于猫眼电影数据和SVR回归器的电影票房预测系统, 包括数据爬取, 特征分析以及数据预测》+源代码+文档说明

import pandas as pd import numpy as np import matplotlib.pyplot as plt import sys from xlrd import open_workbook # xlrd用于读取xld from texttable import Texttable import seaborn as sns import re #票房Top 10 def office_top_10(df): data1=pd.DataFrame(df,columns=['名称','上映时间', '类型','评分','累计票房']) data1['累计票房'] = data1['累计票房'].apply(int) data1 = data1.sort_values(by='累计票房', ascending=False)[0:10] tb=Texttable() #表格展示 tb.header(data1.columns.to_numpy()) tb.add_rows(data1.values,header=False) print(tb.draw()) #票房影响因素 def office_influence(df): data2 = pd.DataFrame(df,columns = ['评分','首周票房','时长','评分人数','累计票房']) data2['评分人数'] = data2['评分人数'].str.replace('万元',' ') data2['评分人数'] = data2['评分人数'].str.replace('万',' ') data2['时长'] = data2['时长'].str.replace('分钟',' ') data2.dropna(axis=0, how='any', inplace=True)#去掉含空值的行 data2 = data2.astype(float) ax = sns.heatmap(data2.corr(), annot=True, vmax=1, square=True, cmap="Blues") ax.set_ylim(5.0, 0) #电影总类型统计 def countN(column): count = dict() for row in column: row = row.split(',') row = row[:len(row)-1] for ele in row: if ele in count: count[ele] += 1 else: count[ele] = 1 return count def movie_type(df): data3 = pd.DataFrame(df,columns = ['名称', '类型','年份']) #每种类型出现的次数除以总的影片数，以此作为该种类型的频数百分比 genres = pd.Series(countN(data3['类型'])).sort_values() genres_avg = genres / len(data3) genres_avg.plot(kind = 'barh', title = '类型频率') #选取前5种类型，观察它们在这15年间每年的数量与当年影片总数之比的变化 genres_by_year = data3.groupby('年份')['类型'].sum() genres_count = pd.DataFrame([], index = genres_by_year.index, columns = genres.index[len(genres)-5:]) print(len(genres)) for g in genres_count.columns: for y in genres_count.index: genres_count.loc[y,g] = genres_by_year[y].count(g) / len(genres_by_year[y]) genres_count.plot(figsize = (10,6), title = 'Evolution of Movies in 5 Genres') #不同类型影片的票房 #对于某种类型，计算所有该类影片的票房，再除以该类影片的数量 def differ_type(df): data3 = pd.DataFrame(df,columns = ['名称', '类型','年份']) genres = pd.Series(countN(data3['类型'])).sort_values() movies_by_genres = pd.DataFrame(0, index = genres.index, columns = ['累计票房', '评分']) for i in range(len(df)): row = df['类型'][i].split(',') row = row[:len(row)-1] for g in row: movies_by_genres.loc[g, '累计票房'] += df['累计票房'][i] #该类影片的总票房 movies_by_genres.loc[g, '评分'] += df['评分'][i] #该类型影片的总评分 movies_by_genres = movies_by_genres.div(genres.values, axis=0) movies_by_genres.sort_values('累计票房', ascending=False)[['累计票房']].plot( kind = 'bar', title='Average Revenue in Different Genres') return movies_by_genres #导演的票房分布 def director_office(df): df['累计票房'] = df['累计票房'].astype(float) revenue_of_director = df.groupby('演职人员1').累计票房.mean() #平均票房 return revenue_of_director #主演的票房分布 def count_actor(column): count = dict() for row in column: row = row.split(',') for ele in row: if ele in count: count[ele] += 1 else: count[ele] = 1 return count def score_actor(df): # movies_noani = df[~df['类型'].str.contains('动画', regex=False)].reset_index(drop = 'True') movies_noani = df movies_noani['演职人员2'] = movies_noani['演职人员2'].replace(' ',' ')+','+movies_noani['演职人员3'].replace(' ',' ') actors = pd.Series(count_actor(movies_noani['演职人员2'])).sort_values() movies_by_actors = pd.DataFrame(0, index = actors.index, columns = ['累计票房', '评分']) #按不同权重统计演员的票房： r4 = [0.4, 0.3, 0.2, 0.1] #如果有4位主演，按此加权，以下类似 r3 = [0.4, 0.3, 0.3] r2 = [0.6, 0.4] r1 = [1] r = [r1, r2, r3, r4] for i in range(len(movies_noani)): actorlist = movies_noani['演职人员2'][i].split(',')[0:2] for j in range(len(actorlist)): movies_by_actors.loc[actorlist[j], '累计票房'] += movies_noani['累计票房'][i] * r[len(actorlist)-1][j] #一个演员的总票房 movies_by_actors.loc[actorlist[j], '评分'] += movies_noani['评分'][i] * r[len(actorlist)-1][j] #一个演员的总评分 movies_by_actors = movies_by_actors.div(actors.values, axis=0) #求平均值 return movies_by_actors #档期分布 def month_movie(df): df['month'] = pd.to_datetime(df['上映时间']).apply(lambda x: x.month) revenue_month = df.groupby('month')['累计票房'].sum() / df.groupby('month').size() revenue_month.plot(kind='bar', title='Average Revenue per Month') return revenue_month ##main函数 # if __name__ == "__main__": # #数据读取并预处理 # data = pd.read_excel(file) # df = pd.DataFrame(data) # df = df[(df.累计票房!='暂无')] # df['年份'] = pd.to_datetime(df['上映时间']).apply(lambda x: x.year) # df.dropna(axis=0, how='any', inplace=True)#去掉含空值的行 # df = df[~df.isin(['暂无'])] # df.reset_index(drop=True, inplace=True)#重新排序 # file = 'd:\\movie.xls' # office_top_10(df) # office_influence(df) # movie_type(df) # differ_type(df) # revenue_of_director = director_office(df) # revenue_of_director.sort_values().tail(10).plot(kind = 'barh', title = 'Directors with Top Revenue') # actor = score_actor(df) # actor['累计票房'].sort_values().tail(10).plot(kind = 'barh') # month_movie(df) # plt.rcParams['font.sans-serif']=['SimHei'] #显示中文标签 # plt.rcParams['font.serif'] = ['KaiTi'] # plt.rcParams['axes.unicode_minus'] = False # plt.show()