Client background: A global expert in data capture, categorisation and structure sectors

 

Improved data categorization accuracy and critical behavioural insights

Business Challenge

  • To generate insights for consumer transaction data like consumer’s credit, earning and spending behaviors

  • Identifying preferred customer channel and devising personalized marketing strategies

  • Build predictive analytics around Propensity to spend, Propensity to pay a loan, Propensity to default, Suburb level income and spending levels and patterns. Facilitate better results than the current categorization engine

  • Categorizing Financial Transaction data of its customers achieve maximum accuracy level (match rate, match accuracy and false positives)

Business Solution

  • Performed exploratory analysis on data

  • Utilized ML and AI to come up with Behavioral Insights/ KPIs

  • Used clustering, Regression / Time Series and Association Mining for model building, based on KPIs ( such as Propensity to Pay). Leveraged classification technique using Recurrent Neural Network (RNN), Naive Bayes Classifier and logistic regression for data categorization

  • Created visualizations for exploratory data as well as inferences

Business Benefits

  • Generated behavioral Insights based on KPIs such as insights covering Customer’s Propensity to Spend or Customer Segmentation.

  • Identified main channels of transactions (ATM withdrawal, credit/debit card payments, using Classification algorithms)

  • Rolled out personalized marketing; product cross-selling based on customer segmentation

  • Unearthed correlations in data and produced accurate output as to the categorisation group of data; improved accuracy levels

Environment

  • Python / R

  • Tableau

  • JSON/CSV for data.