当前位置:网站首页>Machine learning | prediction and analysis of customer purchase probability based on machine learning recommendation system

Machine learning | prediction and analysis of customer purchase probability based on machine learning recommendation system

2020-12-07 16:09:15 itread01

I wrote an article last week “ Based on machine learning, the prediction and analysis of customer purchase probability in bank telemarketing ”, It was the first attempt to validate the case as a predictive analysis of customer purchase possibilities . Today is the second verification case of customer purchase possibility prediction analysis based on machine learning : Recommendation system .

Recommendation system

Based on the heat recommendation : By experts or a certain period of product sales or the main push products , Make a leaderboard , Recommend according to the leaderboard without user data

Recommendation based on user characteristics : Through historical data , By the algorithm ( Machine learning ) Make recommendations based on user characteristics , In the case that the user data can fill in some basic information

Recommendation based on knowledge : By user request , Higher returns than if needed , Need low-risk products , Screening products in the database , Then recommend it

Based on content recommendation : Through products that users have already purchased , Recommend similar items , The content here is similar to that provided by professionals

Collaborative filtering recommendation : Through the algorithm directly calculate the item acquaintance degree , Note that the similarity of items here is not 2 Item content , It's a purchase to determine , Like buying A( Cell phones ) Users of , Most of them have bought B( Mobile phone case ), The algorithm calculates A And B Is similar to that of

Some machine learning recommends : By using some machine learning algorithms , For example FM( Factorizer ), Deep learning ,deepfm etc.

For new users, one of three is usually chosen For old users , Usually, a variety of algorithms are used , Then weighted to get the best recommendation list

Verification method

The main thing here is 3 Three methods to verify , Collaborative filtering ,FM( Factorizer ),deepFM.

Collaborative filtering algorithm

Calculate the similarity of all items ( Note that the content is not similar )

Recommended method :  It's enough for users ( Or evaluated ) The article query acquaintance degree of , Then weighted recommendation

A simple example

Finally recommend ( Left to right )

Based on collaborative filtering display Yes user=‘1’

Python, Printing results , The movies I've seen here are just before printing 5, Recommended movies only recommend weighted before 10

------WATCHED MOVIES--------

1193 "One Flew Over the Cuckoo's Nest (1975)" 'Drama'

2355 "Bug's Life, A (1998)" "Animation|Children's|Comedy"

1287 'Ben-Hur (1959)' 'Action|Adventure|Drama'

2804 'Christmas Story, A (1983)' 'Comedy|Drama'

595 'Beauty and the Beast (1991)' "Animation|Children's|Musical"

------RECOMMEND MOVIES--------

1196 'Star Wars: Episode V - The Empire Strikes Back (1980)' 'Action|Adventure|Drama|Sci-Fi|War']]

1265 'Groundhog Day (1993)' 'Comedy|Romance' 364 'Lion King, The (1994)' "Animation|Children's|Musical"

260 'Star Wars: Episode IV - A New Hope (1977)' 'Action|Adventure|Fantasy|Sci-Fi'

2571 'Matrix, The (1999)' 'Action|Sci-Fi|Thriller'

2716 'Ghostbusters (1984)' 'Comedy|Horror'

1022 'Cinderella (1950)' "Animation|Children's|Musical"

318 'Shawshank Redemption, The (1994)' 'Drama'

1282 'Fantasia (1940)' "Animation|Children's|Musical"

1580 'Men in Black (1997)' 'Action|Adventure|Comedy|Sci-Fi'

Factorization algorithms

The rating matrix is a matrix that reflects the user's preference for the item , Here's the picture

Factorizer an algorithm to complete the scoring matrix ( All red algorithm ), Then recommend according to the score ( For users 1 Recommended for items 3> goods 6> goods 1> goods 4> goods 5)

The algorithm explains : Find out by a known score P and Q,K For hidden features , You can set different values

The effect of factorization algorithm is as follows

------RECOMMEND MOVIES--------

318 'Shawshank Redemption, The (1994)'

858 'Godfather, The (1972)'

1198 'Raiders of the Lost Ark (1981)'

50 'Usual Suspects, The (1995)'

2858 'American Beauty (1999)'

912 'Casablanca (1942)'

593 'Silence of the Lambs, The (1991)'

750 'Dr. Strangelove or: How I Learned to Stop Worrying and Love the Bomb (1963)'

908 'North by Northwest (1959)'

1221 'Godfather: Part II, The (1974)'

deepFM Algorithm

deepFM It's a combination FM Deep learning algorithms and algorithms , The structure is as follows

deepFM The effect of the algorithm is as follows

------RECOMMEND MOVIES--------

593 'Silence of the Lambs, The (1991)'

1617 'L.A. Confidential (1997)'

1233 'Boat, The (Das Boot) (1981)'

318 'Shawshank Redemption, The (1994)'

1198 'Raiders of the Lost Ark (1981)'

858 'Godfather, The (1972)'

733 'Rock, The (1996)'

1276 'Cool Hand Luke (1967)'

2571 'Matrix, The (1999)'

953 "It's a Wonderful Life (1946)"

Model evaluation index

  1. User satisfaction
  2. Prediction accuracy
  3. Coverage
  4. Diversity
  5. Novelty
  6. Surprise degree
  7. Trust degree
  8. Real time
  9. Robustness
  10. Business goals

The goal of the offline experiment

Maximize , Prediction accuracy

In the case of meeting certain requirements

such as :

  • Coverage >60%
  • Diversity >30%
  • Novelty >10%

The effectiveness evaluation needs to be supplemented ......

Summary of recommendation system algorithm

  1. Many algorithms are suitable for different scenarios .
  2. Mixed recommendation , Increase the buying rate . For example, users buy mobile phones , Similar in content , Recommend other mobile phones ; By collaborative filtering , People who recommend mobile phone cases and so on to buy mobile phones will probably buy the size of ; According to the algorithm of score estimation , Beer that may be purchased according to the user's recommendation , sartorial . In limited recommended locations , Mixed recommendation , Improve the success rate .

All I saw was true love , Pay attention when you have time , Give me a comment O(∩_

版权声明
本文为[itread01]所创,转载请带上原文链接,感谢
https://chowdera.com/2020/12/20201207160712762j.html