The data used for prediction is analysed first like what happened, why it is happened? Then it is monitored for what is happening now? After that predictive analytics is applied to find out what will happen in future and then a respective action for the prediction is applied. This whole process takes place with respect to time.
analytics allows the organization to become proactive, forward looking anticipating outcomes and behaviour based on data and not on assumptions Predictive Analytics is a technique which helps to predict the unknown future events using many techniques like data mining, modelling, statistics, machine learning, artificial intelligence etc., The data used for predictive analytics can be structured or unstructured. Structured data can be readily used for prediction whereas unstructured data needs to be transformed as a structured data through modelling techniques and then used for prediction.
Our Predictive Analytics will address the sectors like Treasury, HR, Supply chain etc.,
Pain Points of Addressed Sectors
• Corporate accounts often are in multiple currencies and across several geographies, these require complex interactions to be able to view and monitor all the account and transaction information
• New regulations allow treasurers to view and monitor all of their bank accounts on a single platform
• Search on unstructured data (local files & folders)
• Identification of Suppliers
• Integrated view of supplier performance
• Lack of indexation of supply chain and transaction data
• Supply chain risk prediction
EXTENDED USE CASE- PREDICTIVE ANALYTICS
PREDICT OTHER RISKS
q Market Risk
q Regulatory Risk
q Forex Risk
q Supply Chain Risk
q Supplier Scores
q Supplier Tiers
NON FINANCIAL ANALYSIS
q Stream the social data with respect to the bank
q Do sentiment analysis on the data that enables better decision making
Supply Chain Disruptions-Natural Factors
• HR Analytics help organizations remain competitive in several aspects beginning with acquiring top talent and retaining them.
• IT helps us provide evidence-based advice on how to drive the business from a people perspective
• It also enables the HR initiatives to be pursued by line-of-business leaders to help them reach business targets
• Does the value of Turnover mentioned as a specific number without any meaningful insight given in the monthly Dashboards make any sense to the business leaders?
• Front-line leaders simply do not have the time to pore over numerous metrics prior to making any informed decisions. They expect data represented in simple format that enables them make better decisions
Example: Bank BPO hires ~ 30,000 Customer Service Reps (CSRs) a year to work in their call centers. Approximately 33% made it through the 12 weeks of training, but failed the required Series 7 Exam at the end of the training. Need to find Finding the right variables that could uncover the “raw talent profile”, and apply Statistical analysis to correlate this with top/bottom CSR performance.
WORKFORCE DIVERSITY, MOBILITY & EQUALITY
• Are we paying men and women the same?
• How much should we offer new hires to balance compensation?
TEAM PRODUCTIVITY, QUALITY & ENGAGEMENT
• Are my recruiters and team members happy at their job?
• Are hiring managers pleased with the service they are receiving?
EMPLOYEE REFERRALS AND CUSTOMER ADVOCACY
• Did we get back to that referred candidate?
• Are we asking for referrals across the business?
REVENUE AND BRAND PROTECTION
• How many customers are we saying NO to as applicants?
Is our candidate experience losing us customers?
CASE STUDY FOR BARCLEYS
Risk Analytics for HR Teams : Attrition Prediction
Data Preparation & Cleaning
• Categories for age, years in service, salary, grade, etc. Created.
• Missing performance ratings replaced with root mean squared values.
• Outliers and invalid data points removed.
• Variable called Attrition created (value = 0 or 1).
(Phase 1) Probability of Attrition
• Dataset split 70:30, for use in model building and testing respectively.
• 3000+ different samples created, each to train the model in a unique iteration.
• This model only predicted whether employee would attrite or not.
• Prediction accuracy achieved = 74%.
(Phase 2) Risk Modelling
• Phase 2 calculated each employees’ risk of attrition.
• Risk categories: High (score > 90%), Medium (90% > score > 60%), Low (60% > score > 20%), Safe (score < 20%).
• Also computed reasons for employee leaving – manager, better career, education, etc.
• Model’s accuracy = 84%.
Final Model and Visualization
• Final model would include tenure of employee at organization and Manager.
• Model would forecast # of months until employee leaves.
• A UI will enable HR to feed in raw data, run the model and generate year-, department-, grade- and tenure-wise reports for attrition.