Analysis Of Factors Affecting Photo Download Rate Based On Statistical Methods

Photos serve as a carrier of information and are used by people in various scenarios.In the Internet era,the speed and cost of delivering photos are minimal,and people often use social software to share their photos in everyday life.The contents of the photos are varied,one of which is sports photography.The short-distance marathon race,as a kind of game with low difficulty and a large number of participants,has gradually received the attention of many amateur runners,and many people participate in it.After the competition,many runners wanted to get photos of participating in the competition.Many photography companies seized the opportunity and provided corresponding services.In order to provide players with more and better photos,and increase the profits of enterprises,how to increase the download rate of photos is the company's most concerned issue.In order to make the investigation and research more targeted,first of all,a questionnaire survey was conducted on amateurs participating in the marathon to determine the core user groups and their characteristics,and to make judgments on the market trends of sports photography.Through the survey of users,it is found that the most important requirement of the user for photos is that the subject of the photo is the user himself and is clear and can obtain his own photos as fast as possible after the game.Therefore,the main factors affecting the photo download rate are divided into two parts: the aesthetic quality of the photos and the efficiency of image uploading.On the basis of this,the indicators such as the sharpness,saturation,brightness,image color,and subject position information of the photograph are presented to reflect the aesthetic quality of the image.And the statistical model was used to analyze the effect of the aesthetic quality data of the study photos on the download rate of the photos.Finally,logarithmic probability regression,linear discriminant analysis,decision tree model,and random forest model were compared through experiments,and an accuracy rate of 78% was obtained.The classification model shows that the final selected index system can well identify the aesthetic quality of the image.And can identify the image with poor quality,and combine the data to give retouching suggestions.This article applies the statistical analysis method to this new scenario and actual business,and obtains good results.It also has practical reference value for the enterprise.Combining the company's recommendation algorithm can provide users with more high-quality images,satisfying the users.Under the premise,get more profits.