PwC’s GR D&A's Intelligent Forecasting Platform (IFP) is an advanced big data AI based analytics platform designed to process limitless data volumes of time series with cutting-edge algorithmic capability.
It empowers organisations to make data-driven strategic and operational decisions through prediction model lifecycle process automation, while it streamlines planning and reduces uncertainty.
Highlights
- Integrated with any ERP and/or CRM platforms and any IoT Data Processing device
- Minimizes forecast generation and production using the most appropriate predictive algorithm for any given time series
- Monitors the end to end prediction model lifecycle process via Business Intelligence interactive dashboards, where a user can visualize, analyze and interpret the model outputs and the produced forecasts.
Key features
- Web Interface: Initial setup and any kind of interaction is done via a user-friendly interface allowing users to fully exploit the assessed features.
- Pre-processing: Anomaly detection and normalising, timeseries categorisation
- Automated model development and selection: A big data platform enabling a variety of different predictive algorithms (traditional statistics and advanced ML) and limitless time series data volumes and validating in terms of forecasting accuracy.
- Ability to select models (automatically/manually): End user is able to either automatically or manually select a model of preference for a certain business or combination of segments
- Analysis of exogenous variables: IFP is capable of detecting the most impactful exogenous variables and visualise their contributions to the final forecasts.
- BI management reporting: IFP is capable of creating standard and custom Business Intelligence reports of model outputs based on business requirements.
- Vendor agnostic: IFP can be deployed both on premises and on any public cloud platform.
Benefits
- Analytics democratisation via user-friendly web interface
- Cutting-edge big data forecasting engine providing increased accuracy in minimum lead times
- Automated validation and selection process of advanced statistical and ML prediction algorithms
- Ability to manually select a model for a specific segment or sub-segment adds extra flexibility
- Transparent model validation dashboards
- Ability to be integrated with any ERP and/or CRM platforms and or/and any IoT data processing device
- Ability to handle a growing amount of work presenting limitless scalability
- Low cost and cutting edge big data technology solution
- Value creation utilising the outputs of post model BI management reporting dashboards
IFP enables automated, rapid business insights in the whole spectrum of decision making through PwC GR Analytics as a Service (AaaS) platform.
Business Benefits
- 90% of processing time savings on time series forecasting generation - this means you spend more qualitative time on analysis
- 25% improving accuracy of time series forecasts - this mean operational savings and better planning
- 10% of resource utilization on producing forecasts - this means improvements on resource utilisation
- 20 seconds the execution of 100 number of time series and 5 number of parameters - this means more time to run any experiment you like with any type of data and hence increase your innovation state
Interactive BI dashboards fully customizable to your business requirements ready to meet your operational challenges. IFP tool can be run by either your technical team(s) or be a fully outsourced under a Flexible Platform as a Service (PaaS) commercial package.
How it works
In order to produce accurate forecasts and harness data-driven insight, we follow a comprehensive and coherent approach consisting of 4 steps.
- Data quality validation
- Pre-processing steps
- Add client-specific business intelligence
- Generate consensus forecast
Data quality validation
Perform data quality checks and verify desired data quantity in order to deliver a base forecast.
Pre-processing steps
Perform anomaly detection and smoothing for a client dataset as well as timeseries categorisation.
Add client-specific business intelligence
Identify market or client-specific factors that need to be taken into account to improve forecast accuracy (indicatively: customer segments, product IDs, SKU combinations, competition, temperature, promo calendar, covid related, macroeconomic factors, etc.).
Generate consensus forecast
Produce the best possible forecast per predetermined segment combination. Provide forecasting validation and business segmentation through a BI dashboarding capability.
We produce an analytical description across all prediction modeling lifecycle phases by leveraging a friendly user interface.
IFP UI provides visualization and transparency of the entire prediction model lifecycle and forecast outputs
During the Modelling phase, the production of multiple forecasting models and the most appropriate is implemented
The Pre-Modelling dashboard provides the user with descriptive information on data
In the Post-Modelling dashboard, the user has a dynamic view of the results via an interactive dashboard
Director, Advisory, Data & Analytics Leader, Athens, PwC Greece
Konstantinos Kanistras
Senior Manager, Advisory, Athens, PwC Greece
Manager, Advisory, Athens, PwC Greece
