63
Neural Network Based Player Retention Prediction in Free to Play Games Amr Koura First Examiner: Prof. Dr. Christian Bauckhage Second Examiner: Prof. Dr. Stefan Wrobel Advisor: Rafet Sifa 1

Neural Network Based Player Retention Prediction in Free to Play Games

Embed Size (px)

Citation preview

Page 1: Neural Network Based Player Retention Prediction in Free to Play Games

Neural Network Based Player Retention Prediction in Free to Play Games

Amr KouraFirst Examiner: Prof. Dr. Christian Bauckhage

Second Examiner: Prof. Dr. Stefan WrobelAdvisor: Rafet Sifa 1

Page 2: Neural Network Based Player Retention Prediction in Free to Play Games

Content

• Introduction• Related Work• Solution Approach• Future Work• Conclusion

2

Page 3: Neural Network Based Player Retention Prediction in Free to Play Games

https://www.statista.com/statistics/234649/percentage-of-us-population-that-play-mobile-games/ 3

Page 4: Neural Network Based Player Retention Prediction in Free to Play Games

https://wallpaperscraft.com/download/call_of_duty_modern_warfare_3_france_eiffel_tower_soldiers_battle_tanks_19772/1920x12004

Page 5: Neural Network Based Player Retention Prediction in Free to Play Games

Call of Duty MW3

1.5 million peoplenight #1

6.5 million copiessold on launch day

775$ millionin first 5 days

1$ billionin 16 days

https://en.wikipedia.org/wiki/Call_of_Duty:_Modern_Warfare_3

5

Page 6: Neural Network Based Player Retention Prediction in Free to Play Games

Freemium

Free-to-play game http://www.techlamb.com/2014/11/candy-crush-soda-saga-for-pc-download.html 6

Page 7: Neural Network Based Player Retention Prediction in Free to Play Games

Freemium

https://www.pinterest.com/pin/508554982898643116/ 7

Page 8: Neural Network Based Player Retention Prediction in Free to Play Games

“Who will leave the game in the future?”

8

Page 9: Neural Network Based Player Retention Prediction in Free to Play Games

Defining Problem ?

9

Page 10: Neural Network Based Player Retention Prediction in Free to Play Games

The Big Picture

Game Player

Player Activities

Churn Prediction

10

Page 11: Neural Network Based Player Retention Prediction in Free to Play Games

The Big Picture

Game Player

Player Activities

Churn Prediction

11

Page 12: Neural Network Based Player Retention Prediction in Free to Play Games

The Big Picture

Game Player

Player Activities

Churn Prediction

12

Page 13: Neural Network Based Player Retention Prediction in Free to Play Games

The Big Picture

Game Player

Player Activities

Churn/Retention Prediction

13

Page 14: Neural Network Based Player Retention Prediction in Free to Play Games

Related Work

14

Page 15: Neural Network Based Player Retention Prediction in Free to Play Games

Related WorkHadiji et al., 2014 studied the churn prediction over five free-to-play games , with several machine learning algorithms.

Decision Tree achieved best prediction performance.

15

Page 16: Neural Network Based Player Retention Prediction in Free to Play Games

Related WorkSifa et al., 2015 studied player purchase decision in mobile free-to-play games over 100,000 players.

16

Page 17: Neural Network Based Player Retention Prediction in Free to Play Games

Related WorkDrachen et al., 2016 studied the player retention in the free-to-play mobile games.

17

Page 18: Neural Network Based Player Retention Prediction in Free to Play Games

Solution Approach

18

Page 19: Neural Network Based Player Retention Prediction in Free to Play Games

Classical Machine Learning

Algorithms

Deep learning Approach

19

Page 20: Neural Network Based Player Retention Prediction in Free to Play Games

Classical Machine Learning Algorithms

20

Page 21: Neural Network Based Player Retention Prediction in Free to Play Games

Data setHave historical data with the following features.

Description Field NameDay of event Observation daySum of daily move count Moves CountSum of daily Opponents Active OpponentsSum of daily World Number World NumberSum of daily Skill 1 scores Skill 1Sum of daily Skill 2 scores Skill 2Sum of daily Skill 3 scores Skill 3Sum of daily reached goals Goals reachedSum of daily time in game Time in gameWhich day of week Week DayPlayer country code Country code

21

Page 22: Neural Network Based Player Retention Prediction in Free to Play Games

Data setHave historical data with the following features.

Description Field NameDay of event Observation daySum of daily move count Moves CountSum of daily Opponents Active OpponentsSum of daily World Number World NumberSum of daily Skill 1 scores Skill 1Sum of daily Skill 2 scores Skill 2Sum of daily Skill 3 scores Skill 3Sum of daily reached goals Goals reachedSum of daily time in game Time in gameWhich day of week Week DayPlayer country code Country code

11*7=77

Feature/Player

22

Page 23: Neural Network Based Player Retention Prediction in Free to Play Games

Classical Machine Learning● Logstic regression.

● Decision tree

● SVM(Polynomial kernel , RBF kernel)

● Random Forest (100 tree)

23

Page 24: Neural Network Based Player Retention Prediction in Free to Play Games

Classical Machine Learning

24

Page 25: Neural Network Based Player Retention Prediction in Free to Play Games

Deep learning Approach

25

Page 26: Neural Network Based Player Retention Prediction in Free to Play Games

Motivation

26

Page 27: Neural Network Based Player Retention Prediction in Free to Play Games

“To achieve its impressive performance in tasks such as speech perception or object recognition, the brain extracts multiple levels of representation from the sensory input.” Hinton ,2007.

Motivation

27

Page 28: Neural Network Based Player Retention Prediction in Free to Play Games

Motivation

Distributed Representation of Visual Datahttp://www.slideshare.net/WillStanton/deep-learning-with-text-v4

28

Page 29: Neural Network Based Player Retention Prediction in Free to Play Games

Motivation

Shallow neural network vs Deep neural networkhttp://stats.stackexchange.com/questions/182734/what-is-the-difference-between-a-neural-network-and-a-deep-neural-network

29

Page 30: Neural Network Based Player Retention Prediction in Free to Play Games

Feed Forward Neural Network

30

Page 31: Neural Network Based Player Retention Prediction in Free to Play Games

Deep Learning

Input Layer

Output Layer

Hidden Layer

Weights W

Weights W

77 Neurons

200 Neurons

2 Neurons

31

Page 32: Neural Network Based Player Retention Prediction in Free to Play Games

Deep Learning

Input Layer

Output Layer

Hidden Layer

Weights W

Weights W

77 Neurons

RELU

Soft Max

Hint: Learning rate = 0.3 Batch size = 2700 Training Iteration = 1000 32

Page 33: Neural Network Based Player Retention Prediction in Free to Play Games

Deep Learning

33

Page 34: Neural Network Based Player Retention Prediction in Free to Play Games

Deep Learning Supervised pre-training

34

Page 35: Neural Network Based Player Retention Prediction in Free to Play Games

Recurrent Neural Network

35

Page 36: Neural Network Based Player Retention Prediction in Free to Play Games

Motivation

Recurrent Neural Networkhttps://leonardoaraujosantos.gitbooks.io/artificial-inteligence/content/recurrent_neural_networks.html

36

Page 37: Neural Network Based Player Retention Prediction in Free to Play Games

Recurrent Neural Network Pre-processing:Create 3D arrays [#players, #days, #features].

Player 1 Data [7x11]

Player 2 Data [7x11]

Player N Data [7x11]37

0

1

037

Page 38: Neural Network Based Player Retention Prediction in Free to Play Games

Recurrent Neural Network Pre-processing:In case of missing data , we pad the player records with zeros.

3

4

5

6

7

8

9

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .0

. . . . . . . . . . .0000000000

38

Page 39: Neural Network Based Player Retention Prediction in Free to Play Games

Unfolding

Recurrent Neural Network Unfoldinghttp://www.wildml.com/2015/09/recurrent-neural-networks-tutorial-part-1-introduction-to-rnns/

39

Page 40: Neural Network Based Player Retention Prediction in Free to Play Games

Unfolding through time

40

Page 41: Neural Network Based Player Retention Prediction in Free to Play Games

Unfolding through time

41

Page 42: Neural Network Based Player Retention Prediction in Free to Play Games

Unfolding through time

t

t-1

t-2

t-3

42

Page 43: Neural Network Based Player Retention Prediction in Free to Play Games

RNN Modes

RNN Modeshttp://www.wildml.com/2015/09/recurrent-neural-networks-tutorial-part-1-introduction-to-rnns/

43

Page 44: Neural Network Based Player Retention Prediction in Free to Play Games

LSTMLSTM solves one RNN learning problem 'vanishing exploding gradients'

Gated LSTM Cell https://en.wikipedia.org/wiki/Long_short-term_memory

44

Page 45: Neural Network Based Player Retention Prediction in Free to Play Games

LSTM Network

LSTM Networkhttps://deeplearning4j.org/lstm 45

Page 46: Neural Network Based Player Retention Prediction in Free to Play Games

Bidirectional LSTM

Bidirectional LSTM Networkhttp://www.cs.cmu.edu/afs/cs/user/zhouyu/www/ASRU.pdf 46

Page 47: Neural Network Based Player Retention Prediction in Free to Play Games

Dynamic LSTM

Dynamic LSTM Networkhttps://arxiv.org/pdf/1406.1078.pdf 47

Page 48: Neural Network Based Player Retention Prediction in Free to Play Games

Dynamic Bidirectional LSTM

Accuracy :0.899078983403 Precision :0.827267953923 Recall :0.884941525098 F1-score :0.854869382267 G mean :0.632014710972

48

Page 49: Neural Network Based Player Retention Prediction in Free to Play Games

Evaluation Metric

49

Page 50: Neural Network Based Player Retention Prediction in Free to Play Games

Future Work

50

Page 51: Neural Network Based Player Retention Prediction in Free to Play Games

Predict the player Retention for the next Month rather than next week.

Future Work

51

Page 52: Neural Network Based Player Retention Prediction in Free to Play Games

Sequence-to-sequence learning model rather than sequence-to-label.

Future Work

52

Page 53: Neural Network Based Player Retention Prediction in Free to Play Games

use unsupervised pre-training before learning deep learning model using stacked Autoecoder.

Future Work

53

Page 54: Neural Network Based Player Retention Prediction in Free to Play Games

solve the variable length time series embedding using Autoencoder.

Future Work

Gianniotis et al., 2016 Model-Coupled Autoencoder for Time Series Visualisation 54

Page 55: Neural Network Based Player Retention Prediction in Free to Play Games

Conclusion• Churn prediction analysis is very important task for

game development company.

• This is the first time , the problem solved using

memory based computation.

• RNN is very powerful tool to learn sequence.

• Dynamic Bidirectional LSTM solves the variable

Length time series and gives best learning results.

55

Page 56: Neural Network Based Player Retention Prediction in Free to Play Games

Thank you

56

Page 57: Neural Network Based Player Retention Prediction in Free to Play Games

Appendix

57

Page 58: Neural Network Based Player Retention Prediction in Free to Play Games

BackPropagation

58

Page 59: Neural Network Based Player Retention Prediction in Free to Play Games

Back-propagation through time(BPTT)

59

Page 60: Neural Network Based Player Retention Prediction in Free to Play Games

BPTT

BPTThttp://www.wildml.com/2015/10/recurrent-neural-networks-tutorial-part-3-backpropagation-through-time-and-vanishing-gradients/

60

Page 61: Neural Network Based Player Retention Prediction in Free to Play Games

Gated LSTM Cell

61

Page 62: Neural Network Based Player Retention Prediction in Free to Play Games

Bidirectional LSTM

62

Page 63: Neural Network Based Player Retention Prediction in Free to Play Games

Thank you

63