Financial Transactions Fraud Detection

0 reviews

This solution is a deep learning-based approach to learn and understand the patterns in financial transaction data. It aims at learning the normal behavior patterns of the transactions during the training process using a Restricted Boltzmann Machine algorithm. Once trained, the model can identify abnormal patterns of transactions, thereby classifying them as anomalous.

Newspaper Customer Churn Prediction

0 reviews

Customer churn refers to the loss of existing clients or customers. This solution identifies newspaper customers who are more likely to discontinue their current subscription. During the training stage, the solution automatically conducts feature interaction on the training data and selects a subset of features based on feature importance. It then trains multiple models and identifies the best performing model. This model is then selected for prediction on new data.

Restaurant Reviews Topic Extraction

0 reviews

This solution identifies the various aspects a reviewer is mentioning when providing a review for any restaurant business. This can help businesses easily identify which are its most prominent aspects (e.g. price, ambience, taste, quality etc.) which are getting reviewed and what are the associated opinions about them. They can then improve on these aspects to provide a superior customer experience.

Cloud Network Cost Forecasting

0 reviews

Cloud Network Cost Forecasting generates 24 hours forward forecast of network cost using historical data. This solution will help businesses to better optimize their on-cloud network infrastructure and foresee their cost fluctuations. It uses ensemble ML algorithms with automatic model selection algorithms. This solution provides consistent and better results due to its ensemble learning approach. This solution performs automated model selection to apply the right model based on the input data.

Topic Extraction from Mobile App Reviews

0 reviews

The solution categorizes mobile application reviews into four prominent categories based on user reviews: User Experience, Safety and Security, Functional stability, and Ease of use. Deep learning based simple transformer learners are used to classify the review into one of the above categories.

Active Learning for Text Classification

0 reviews

Active Learning for Text Classification trains a text classification model using a small corpus of training data and provides the most appropriate samples from a huge corpus of unlabeled data to be annotated in order to improve the model accuracy significantly. Using Active Learning this algorithm helps in identifying the most effective data sample to be tagged first thus reducing the time and effort to build a usable Machine learning model.

Broadband Customer Churn Prediction

0 reviews

Customer churn refers to the loss of existing clients or customers. This solution identifies broadband customers who are more likely to discontinue their current broadband service provider. During the training stage, the solution automatically conducts feature interaction on the training data and selects a subset of features based on feature importance. It then trains multiple models and identifies the best performing model. This model is then selected for prediction on new data

Customer Complaint Ticket Classification

0 reviews

A high frequency of issues can generate an overwhelming number of customer complaint tickets and incorrect delegation to teams to handle them. This leads to a spike in MTTR (mean time taken to resolve) and a dip in FCR (First Call Resolution). The solution mitigates these issues by training a multi-factor ML model that considers factors like ticket impact, urgency, priority, issue description and other features to predict the most relevant group to resolve a ticket. A pool of models is run through data to select the most generalizable model for the ticket classification task.

Airline Tweets Sentiment Analyzer

0 reviews

This solution classifies tweets mentioning airline travel into positive, neutral and negative sentiments. It uses text analysis, natural language processing, machine learning techniques to predict sentiment classes for tweets. It automates the manual effort to analyze airline travel related tweets and helps generate faster actionable insights around airline services.