
Information
- Date : 2016
- Languages : NodeJS,Python
- Tech Stack : Apache Spark
- Client : SimpleML, SRL.
Project Brief:
For the machine learning startup SimpleML, we developed a fully-automated data science platform that finds the most appropriate machine learning algorithm, optimises its hyperparameters and exports the model to popular programming language for offline evaluation. Main highlights of the platform:
- Scientist or not, all love it alike! Non-experts can leave all the complexities of machine learning to SimpleML's fully-automated platform. Data Scientists may choose which hyperparameters to tune using bayesian optimization or grid search.
- Export and run everywhere: Export machine learning models to your favorite programming language for offline evaluation. Or share the link and run your model from the browser.
- Internet of Things ready: Lightweight model generation for Arduino, Raspberry Pi and other microcontrollers. Aditionally, models can be evaluated online using our MQTT and CoAP webservices.

Marketing
Predict churn, lifetime value, cross/up-selling opportunities, wallet share, reactivation likelihood. Optimize product mix, channels, customer segmentation and discount targeting.

Sales and Logistics
Forecast demand, prioritize leads, optimize inventory, merchandizing and product layout in stores, analyze shrinkage and warranties, predict new store locations and in-store traffic patterns.

Risk, Fraud and Insurance
Predict credit, treasury or currency risk, credit card or transaction fraud, money laundering, liability recovery. Predict insurance claims, optimize agent performance, claims handling and investments.

Internet of Everything
Detect anomalies in sensor measurements, combine sensor readings, search for patterns in sensor data streams. Forecast sensor measurements and predict external events based on historical data.

Customer Service
Forecast call center volume for staff rostering, determine waiting times and call duration. Optimize call routing and which messages are shown on operators’ screen. Predict solutions to common tech issues.

Human Resources
Predict which employees are likely to leave, screen resumes, manage talent, predict employee happiness and engagement, recommend specific training based on performance review data. Workforce management.

Sentiment Analysis
Determine whether an attitude towards a topic or product is positive, negative or neutral. Identify and categorize opinions expressed in text. Classify text in multiple categories (ie. spam or ham).

Other Use Cases
Drug/chemical discovery and analysis, clinical trial design, predict drug demand, survival analysis, prescription compliance, quality management, electrical grid distribution, aircraft scheduling, ...