FeatureStore Parameters for GCP:

featurestore_id

entity_type_id

description

entity_id

finscore_fs

score_features

TBD

Everything Else - Need to align data format

Link - FinScore API Docs

In general, the response depends on the validity of the body content passed to it. For the purposes of FeatureStore, only valid responses results are detailed below. These results will be consumed as features.

Features will be converted and ingested as STRING type.

GENERIC RESPONSE DETAILS

In Response Tab, the response status must be 200. See the Response Codes below:

  • 0000 OK - Successful Operation

  • 1000 OK - Score not Available

  • 1111 Error - General Error

In case there is an error in authentication, the server must return a response code 401 (Unauthorized). If the token is expired, the possible response message is “Token Expired. The token has exceeded its lifetime of 3600 seconds.”GENERIC RESPONSE DETAILS

Response codes will be kept as a feature with feature_id: code

GENERIC SCORE - score_labels

The Finscore’s generic telco scores are statistical models built on various MNO data sources – Data and Voice usage, Top-up Patterns, Location and Device Data and many more. Derived by the most advanced machine learning algorithms, the scores have significant predictive power and can be used as either standalone models or in combination with exiting internal Social-Demographic and/or Credit Bureau models.

{
  "code": "0000",
  "requestid": "338791112",
  "transactionid": "4036a2f8ff1d4386928d98d22e9a06c5",
  "result": "ok",
  "scores": {
      "10": [
        ["score",560],
        ["band",6]
      ]
    }
}

Valid SMART Prepaid:

Response Result

Description

feature_id

Atomic

Type

Score Range

300 to 600

score

Yes

INT

Score Band

1 to 10

band

Yes

INT

  • Valid SMART Postpaid:

Response Result

Description

feature_id

Atomic

Type

Score Range

700 to 999

score

Yes

INT

Score Band

21, 22, 23

band

Yes

INT

GEO - LOCATION FRAUD FLAG

Geo Location Fraud Flag derived from FinScore’s ability to match applicants declared living address versus the real geo – location detected by the mobile operator.- The Finscore’s generic telco scores are statistical models built on various MNO data sources – Data and Voice usage, Top-up Patterns, Location and Device Data and many more. Derived by the most advanced machine learning algorithms, the scores have significant predictive power and can be used as either standalone models or in combination with exiting internal Social-Demographic and/or Credit Bureau models.

{
  "code": "0000",
  "address_check": {
    "20": "000",
    "21": "000"
  },
  "requestid": "338791112",
  "result": "ok",
  "transactionid": "780FBFABC3744B4B8C59F7ADFA9BC19E"
}
  • Valid Address:

Response Result

Description

feature_id

Atomic

Type

(000)

  • 0 First Value is the Client's Location in Daytime.

  • 0 Second Value is the Client's Location in Nighttime.

  • 0 Third Value is the Client's Location in Weekend.

address_check

Yes

STRING

SOCIAL CIRCLE FRAUD FLAG

Social Circle Fraud Flag leverages on the Telco Data and compare the applicant’s guarantor number with the applicant’s most often called numbers.

{
  "code": "0000",
  "msisdn_check": {
    "30": "(000000000,000000000,00000,00000,00000)",
    "31": "(000000000,000000000,00000,00000,00000)"
  },
  "requestid": "338791112",
  "result": "ok",
  "transactionid": "780FBFABC3744B4B8C59F7ADFA9BC19E"
}
  • Valid Mobile Number:

Response Result

Description

feature_id

Atomic

Type

(000000000,000000000,00000,00000,00000)

  • 000000000 First Column is the Client's Top Voice Call

  • 000000000 Second Column is the Client's Top SMS.

  • 00000 Third Column is the Client's Top Recipient / Sender of Pasaload for 3 months.

  • 00000 Fourth Column is the Client's Top Recipient / Sender of Pasaload for 6 months.

  • 00000 Fifth Column is the Client's Top Recipient / Sender of Pasaload for 12 months.

msisdn_check

STRING

CUSTOMER RECORD PROFILE

Customer Record Profile is a product that provides valuable insights to the Partner without revealing the actual sensitive data of the Customer. The Customer Profile Record has a lot in common with a typical Credit Bureau Record, but is composed solely of Telco data.

{
  "code": "0000",
  "customer_profile": [
    ["TimeWithSmart","Established"],
    ["SubscriptionType","Smart Prepaid"],
    ["SmsUsage","Lowest"],
    ["TopUp","Mid"],
    ["MobileDataUsage","High"],
    ["AvgDailyActivities","Low"]
  ],
  "requestid": "3234345555",
  "result": "ok",
  "transactionid": "afcfd32fedf24360a61cd911e5ccbc4f"
}
  • Valid Mobile Number:

Response Result

Description

Atomic

Type

Score Details will display

  • TimeWithSmart The variable measures the days between today and the day the customer activated the SIM card on.

STRING

  • SubscriptionType The variable shows the customer's subscription plan.

STRING

  • SMSUsage The variable measures the number of going SMS's for the last 3 months.

STRING

  • TopUp The variable measures the airtime amount the customer recharged into his airtime wallet in the past 3 months.

STRING

  • MobileData The variable measures how much mobile data the customer consumed for the last 3 months.

STRING

  • AvgDailyActivities The variable measures the average days between any activity made by the customer in the last 3 months - outgoing calls, sms, mobile data used, airtime top ups and airtime spends.

STRING