Data Plan Backend

Weekly Plan

Google Doc Link

Architecture Diagram

User Parameter Estimation

PDF

Visualization

HTML and D3.js are used to visualize recommended daily usages. In static directory, there is a Python Server to serve static HTML file. To run the HTML visualization,

python static_server.py
Open http://your_ip_address:5000/visual.html

Code Structure

• DataPlan.java
  Class to model the data plan

• User.java
  Class to model the user, including properties like daily weights, alpha, phi and current data plan

• Utilities.java
  Utility functions for general purpose calculations, like calculate monthly utilities, parameters estimation

• UserParamFitter
  A class to estimate user's parameters including daily weights, phi and alpha using Levenberg-Marquardt Algorithm

• DataPlanServer.java
  GRPC Server that provides services like user parameter calibration, utility calculation and so on.

• DataPlanClient.java
  An example GRPC client that sends RPC requests for parameter estimation.

GRPC API

• User Parameter Estimation

syntax = "proto3";

option java_multiple_files = true;
option java_package = "";
option java_outer_classname = "DataPlanProto";
option objc_class_prefix = "DP";


// The data plan service
service DataPlanService {
  rpc GetUserParam (UserParamRequest) returns (UserParamResponse) {} //service to calibrate user parameters
  rpc GetRecommendUsages (RecommendUsagesRequest) returns(UsagesResponse) {}  //service to recommend daily usages
  rpc GetUtility( UtilityRequest ) returns( UtilityResponse ){} //service to compute utitity of given user and dataplan
  rpc GetRecommendedDataPlans( DataPlanRequest ) returns( DataPlanResponse ){} //service to recommend top data plans
  rpc GetRecommendedDataPlans2( DataPlanRequest2 ) returns( DataPlanResponse2 ){} //service to recommend top data plans
  rpc HelloWorld( HWRequest) returns( HWResponse ){}  //hello world for heathcheck
}

// Request for User Parameter Calibration
message UserParamRequest {
  repeated Usage usages = 1;
  double overage = 2;  // if overage > 0, heavy user, otherwise light user
}

// response for User Parameter Calibration
message UserParamResponse {
  UserParams userParams = 1; // calibrated parameters
  UserParamsStd userParamsStd = 2;  // standard deviation for confidence interval

}

// user parameters object
message UserParams {
  double w1 = 1; //weights of each day of week
  double w2 = 2;
  double w3 = 3;
  double w4 = 4;
  double w5 = 5;
  double w6 = 6;
  double w7 = 7;
  double phi = 8; // pi
  double alpha = 9; //alpha
}

// user parameters object
message UserParamsStd{
  double w1 = 1; //weights of each day of week
  double w2 = 2;
  double w3 = 3;
  double w4 = 4;
  double w5 = 5;
  double w6 = 6;
  double w7 = 7;
  double phi = 8; // pi
  double alpha = 9; //alpha
}


// Request for Daily Usages Recommendation
message RecommendUsagesRequest{
  int32 year = 1;
  int32 month = 2;
  UserParams userParams = 3;
  DataPlanMsg dataPlan = 4;
}

// Response of recommended daily usages
message UsagesResponse{
  repeated Usage usages = 1;
}

// Request for Utility Calculation
message UtilityRequest{
  UserParams userParams = 1;
  DataPlanMsg dataPlan = 2;
}

// Response to Utility Calculation Request
message UtilityResponse{
  double utility = 1;
}

//Request for top recommended data plans
message DataPlanRequest{
  UserParams userParams = 1;
}

//response of top data plans
message DataPlanResponse{
  repeated DataPlanMsg dataPlans = 1;
}

//Request for top recommended data plans
message DataPlanRequest2{
  UserParams userParams = 1;
  UserParamsStd userParamsStd = 2;  // standard deviation for confidence interval
}


//response of data plans with detailed max and min utility
message DataPlanResponse2{
  repeated DataPlanMsg2 dataPlans = 1;
}



// Hello World Request
message HWRequest{
  string word = 1;
}
// Hello World Response
message HWResponse{
  string word = 1;
}



//object of daily usage
message Usage {
  int32 year = 2;
  int32 month = 3;
  int32 day = 4;
  double usage = 1;
}

// object of data plan
message DataPlanMsg{
  string name = 1;
  string description = 2;
  double quota = 3;
  double overage = 4;
  double price = 5;

}

// object of data plan
message DataPlanMsg2{
  string name = 1;
  string description = 2;
  double quota = 3;
  double overage = 4;
  double price = 5;
  double utility = 6;
  double max_utility = 7;
  double min_utility = 8;
}

Example Data Plans

Data Plans.CSV

Example Client Request and Response

HelloWorld

word: "Hello from client"
word: "Hello from server!"

A pseudo-data plan for testing

Pseudo data plan created
quota: 1000.0
overage: 0.01
price: 35.0

Response of Calibrated Parameters

Calibrated Params:
w1: 0.042539501125739794
w2: 0.04091936098364235
w3: 0.04309952622742607
w4: 0.04104995612787689
w5: 0.043283407774602364
w6: 2.181604896733802E-11
w7: 0.03910824773889646
phi: 0.007839599730990358
alpha: 0.29769491579217067

Standard Deviation:
w1: 0.0019487065819555071
w2: 0.0020724956857322173
w3: 0.0018883412859305403
w4: 0.0017842397015089533
w5: 0.001675321773520107
w6: 0.0029064101994204306
w7: 0.0022180573115046644
phi: 4.565941540202295E-4
alpha: 0.0040705960946803765

Recommended Top 5 Data Plans

Recommended Data Plans
name: "Verizon Prepaid 7G"
description: "https://www.verizonwireless.com/prepaid"
quota: 7000.0
overage: 0.02
price: 50.0

name: "Verizon Prepaid 3G"
description: "https://www.verizonwireless.com/prepaid"
quota: 3000.0
overage: 0.02
price: 40.0

name: "AT&T Prepaid 6G"
description: "https://www.att.com/prepaid/plans.html"
quota: 6000.0
overage: 0.01
price: 45.0

name: "AT&T Prepaid 1G"
description: "https://www.att.com/prepaid/plans.html"
quota: 1000.0
overage: 0.01
price: 35.0

name: "Verizon Prepaid 10G"
description: "https://www.verizonwireless.com/prepaid"
quota: 10000.0
overage: 0.02
price: 60.0

Resposne of Recommended Daily Usages

Recommended Usages: 
usages {
  usage: 16.921750044338605
  year: 2017
  month: 12
  day: 1
}
usages {
  usage: 2.0997563370221223E-29
  year: 2017
  month: 12
  day: 2
}
usages {
  usage: 33.50831641982961
  year: 2017
  month: 12
  day: 3
}
usages {
  usage: 42.23550960361205
  year: 2017
  month: 12
  day: 4
}
usages {
  usage: 50.632803266779696
  year: 2017
  month: 12
  day: 5
}
usages {
  usage: 44.262066249566885
  year: 2017
  month: 12
  day: 6
}
usages {
  usage: 49.832037799831035
  year: 2017
  month: 12
  day: 7
}
usages {
  usage: 16.921750044338605
  year: 2017
  month: 12
  day: 8
}
usages {
  usage: 2.0997563370221223E-29
  year: 2017
  month: 12
  day: 9
}
usages {
  usage: 33.50831641982961
  year: 2017
  month: 12
  day: 10
}
usages {
  usage: 42.23550960361205
  year: 2017
  month: 12
  day: 11
}
usages {
  usage: 50.632803266779696
  year: 2017
  month: 12
  day: 12
}
usages {
  usage: 44.262066249566885
  year: 2017
  month: 12
  day: 13
}
usages {
  usage: 49.832037799831035
  year: 2017
  month: 12
  day: 14
}
usages {
  usage: 16.921750044338605
  year: 2017
  month: 12
  day: 15
}
usages {
  usage: 2.0997563370221223E-29
  year: 2017
  month: 12
  day: 16
}
usages {
  usage: 33.50831641982961
  year: 2017
  month: 12
  day: 17
}
usages {
  usage: 42.23550960361205
  year: 2017
  month: 12
  day: 18
}
usages {
  usage: 50.632803266779696
  year: 2017
  month: 12
  day: 19
}
usages {
  usage: 44.262066249566885
  year: 2017
  month: 12
  day: 20
}
usages {
  usage: 49.832037799831035
  year: 2017
  month: 12
  day: 21
}
usages {
  usage: 16.921750044338605
  year: 2017
  month: 12
  day: 22
}
usages {
  usage: 2.0997563370221223E-29
  year: 2017
  month: 12
  day: 23
}
usages {
  usage: 33.50831641982961
  year: 2017
  month: 12
  day: 24
}
usages {
  usage: 42.23550960361205
  year: 2017
  month: 12
  day: 25
}
usages {
  usage: 50.632803266779696
  year: 2017
  month: 12
  day: 26
}
usages {
  usage: 44.262066249566885
  year: 2017
  month: 12
  day: 27
}
usages {
  usage: 49.832037799831035
  year: 2017
  month: 12
  day: 28
}
usages {
  usage: 16.921750044338605
  year: 2017
  month: 12
  day: 29
}
usages {
  usage: 2.0997563370221223E-29
  year: 2017
  month: 12
  day: 30
}
usages {
  usage: 33.50831641982961
  year: 2017
  month: 12
  day: 31
}

Resposne of Recommended Daily Usages with max and min utilties

Data Plans and max/min utilities
name: "AT&T Prepaid 1G"
description: "https://www.att.com/prepaid/plans.html"
quota: 1000.0
overage: 0.01
price: 35.0
utility: -30.11239557854294
max_utility: -29.460744399958596
min_utility: -30.252503624766288

name: "AT&T Prepaid 6G"
description: "https://www.att.com/prepaid/plans.html"
quota: 6000.0
overage: 0.01
price: 45.0
utility: -40.112395578542944
max_utility: -39.4607443999586
min_utility: -40.252503624766284

name: "AT&T Prepaid Unlimited"
description: "https://www.att.com/prepaid/plans.html"
quota: 1.7976931348623157E308
overage: 0.01
price: 65.0
utility: -60.112395578542944
max_utility: -59.4607443999586
min_utility: -60.252503624766284

name: "Verizon Prepaid 3G"
description: "https://www.verizonwireless.com/prepaid"
quota: 3000.0
overage: 0.02
price: 40.0
utility: -35.112395578542944
max_utility: -34.4607443999586
min_utility: -35.252503624766284

name: "Verizon Prepaid 7G"
description: "https://www.verizonwireless.com/prepaid"
quota: 7000.0
overage: 0.02
price: 50.0
utility: -45.112395578542944
max_utility: -44.4607443999586
min_utility: -45.252503624766284

name: "Verizon Prepaid 10G"
description: "https://www.verizonwireless.com/prepaid"
quota: 10000.0
overage: 0.02
price: 60.0
utility: -55.112395578542944
max_utility: -54.4607443999586
min_utility: -55.252503624766284

name: "Verizon Prepaid Unlimited"
description: "https://www.verizonwireless.com/prepaid"
quota: 1.7976931348623157E308
overage: 0.02
price: 80.0
utility: -75.11239557854294
max_utility: -74.4607443999586
min_utility: -75.25250362476629